version 0.0.5-alpha
This commit is contained in:
911
previousVersion/0.0.05-alpha/Manifest.toml
Normal file
911
previousVersion/0.0.05-alpha/Manifest.toml
Normal file
@@ -0,0 +1,911 @@
|
||||
# This file is machine-generated - editing it directly is not advised
|
||||
|
||||
julia_version = "1.9.2"
|
||||
manifest_format = "2.0"
|
||||
project_hash = "1d38b0278f78d536c218e3a421dfd88a68063099"
|
||||
|
||||
[[deps.AbstractFFTs]]
|
||||
deps = ["LinearAlgebra"]
|
||||
git-tree-sha1 = "8bc0aaec0ca548eb6cf5f0d7d16351650c1ee956"
|
||||
uuid = "621f4979-c628-5d54-868e-fcf4e3e8185c"
|
||||
version = "1.3.2"
|
||||
weakdeps = ["ChainRulesCore"]
|
||||
|
||||
[deps.AbstractFFTs.extensions]
|
||||
AbstractFFTsChainRulesCoreExt = "ChainRulesCore"
|
||||
|
||||
[[deps.Adapt]]
|
||||
deps = ["LinearAlgebra", "Requires"]
|
||||
git-tree-sha1 = "76289dc51920fdc6e0013c872ba9551d54961c24"
|
||||
uuid = "79e6a3ab-5dfb-504d-930d-738a2a938a0e"
|
||||
version = "3.6.2"
|
||||
weakdeps = ["StaticArrays"]
|
||||
|
||||
[deps.Adapt.extensions]
|
||||
AdaptStaticArraysExt = "StaticArrays"
|
||||
|
||||
[[deps.ArgCheck]]
|
||||
git-tree-sha1 = "a3a402a35a2f7e0b87828ccabbd5ebfbebe356b4"
|
||||
uuid = "dce04be8-c92d-5529-be00-80e4d2c0e197"
|
||||
version = "2.3.0"
|
||||
|
||||
[[deps.ArgTools]]
|
||||
uuid = "0dad84c5-d112-42e6-8d28-ef12dabb789f"
|
||||
version = "1.1.1"
|
||||
|
||||
[[deps.Artifacts]]
|
||||
uuid = "56f22d72-fd6d-98f1-02f0-08ddc0907c33"
|
||||
|
||||
[[deps.Atomix]]
|
||||
deps = ["UnsafeAtomics"]
|
||||
git-tree-sha1 = "c06a868224ecba914baa6942988e2f2aade419be"
|
||||
uuid = "a9b6321e-bd34-4604-b9c9-b65b8de01458"
|
||||
version = "0.1.0"
|
||||
|
||||
[[deps.BFloat16s]]
|
||||
deps = ["LinearAlgebra", "Printf", "Random", "Test"]
|
||||
git-tree-sha1 = "dbf84058d0a8cbbadee18d25cf606934b22d7c66"
|
||||
uuid = "ab4f0b2a-ad5b-11e8-123f-65d77653426b"
|
||||
version = "0.4.2"
|
||||
|
||||
[[deps.BangBang]]
|
||||
deps = ["Compat", "ConstructionBase", "InitialValues", "LinearAlgebra", "Requires", "Setfield", "Tables"]
|
||||
git-tree-sha1 = "e28912ce94077686443433c2800104b061a827ed"
|
||||
uuid = "198e06fe-97b7-11e9-32a5-e1d131e6ad66"
|
||||
version = "0.3.39"
|
||||
|
||||
[deps.BangBang.extensions]
|
||||
BangBangChainRulesCoreExt = "ChainRulesCore"
|
||||
BangBangDataFramesExt = "DataFrames"
|
||||
BangBangStaticArraysExt = "StaticArrays"
|
||||
BangBangStructArraysExt = "StructArrays"
|
||||
BangBangTypedTablesExt = "TypedTables"
|
||||
|
||||
[deps.BangBang.weakdeps]
|
||||
ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
|
||||
DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
|
||||
StaticArrays = "90137ffa-7385-5640-81b9-e52037218182"
|
||||
StructArrays = "09ab397b-f2b6-538f-b94a-2f83cf4a842a"
|
||||
TypedTables = "9d95f2ec-7b3d-5a63-8d20-e2491e220bb9"
|
||||
|
||||
[[deps.Base64]]
|
||||
uuid = "2a0f44e3-6c83-55bd-87e4-b1978d98bd5f"
|
||||
|
||||
[[deps.Baselet]]
|
||||
git-tree-sha1 = "aebf55e6d7795e02ca500a689d326ac979aaf89e"
|
||||
uuid = "9718e550-a3fa-408a-8086-8db961cd8217"
|
||||
version = "0.1.1"
|
||||
|
||||
[[deps.CEnum]]
|
||||
git-tree-sha1 = "eb4cb44a499229b3b8426dcfb5dd85333951ff90"
|
||||
uuid = "fa961155-64e5-5f13-b03f-caf6b980ea82"
|
||||
version = "0.4.2"
|
||||
|
||||
[[deps.CUDA]]
|
||||
deps = ["AbstractFFTs", "Adapt", "BFloat16s", "CEnum", "CUDA_Driver_jll", "CUDA_Runtime_Discovery", "CUDA_Runtime_jll", "CompilerSupportLibraries_jll", "ExprTools", "GPUArrays", "GPUCompiler", "KernelAbstractions", "LLVM", "LazyArtifacts", "Libdl", "LinearAlgebra", "Logging", "Preferences", "Printf", "Random", "Random123", "RandomNumbers", "Reexport", "Requires", "SparseArrays", "SpecialFunctions", "UnsafeAtomicsLLVM"]
|
||||
git-tree-sha1 = "442d989978ed3ff4e174c928ee879dc09d1ef693"
|
||||
uuid = "052768ef-5323-5732-b1bb-66c8b64840ba"
|
||||
version = "4.3.2"
|
||||
|
||||
[[deps.CUDA_Driver_jll]]
|
||||
deps = ["Artifacts", "JLLWrappers", "LazyArtifacts", "Libdl", "Pkg"]
|
||||
git-tree-sha1 = "498f45593f6ddc0adff64a9310bb6710e851781b"
|
||||
uuid = "4ee394cb-3365-5eb0-8335-949819d2adfc"
|
||||
version = "0.5.0+1"
|
||||
|
||||
[[deps.CUDA_Runtime_Discovery]]
|
||||
deps = ["Libdl"]
|
||||
git-tree-sha1 = "bcc4a23cbbd99c8535a5318455dcf0f2546ec536"
|
||||
uuid = "1af6417a-86b4-443c-805f-a4643ffb695f"
|
||||
version = "0.2.2"
|
||||
|
||||
[[deps.CUDA_Runtime_jll]]
|
||||
deps = ["Artifacts", "CUDA_Driver_jll", "JLLWrappers", "LazyArtifacts", "Libdl", "TOML"]
|
||||
git-tree-sha1 = "5248d9c45712e51e27ba9b30eebec65658c6ce29"
|
||||
uuid = "76a88914-d11a-5bdc-97e0-2f5a05c973a2"
|
||||
version = "0.6.0+0"
|
||||
|
||||
[[deps.CUDNN_jll]]
|
||||
deps = ["Artifacts", "CUDA_Runtime_jll", "JLLWrappers", "LazyArtifacts", "Libdl", "TOML"]
|
||||
git-tree-sha1 = "2918fbffb50e3b7a0b9127617587afa76d4276e8"
|
||||
uuid = "62b44479-cb7b-5706-934f-f13b2eb2e645"
|
||||
version = "8.8.1+0"
|
||||
|
||||
[[deps.Calculus]]
|
||||
deps = ["LinearAlgebra"]
|
||||
git-tree-sha1 = "f641eb0a4f00c343bbc32346e1217b86f3ce9dad"
|
||||
uuid = "49dc2e85-a5d0-5ad3-a950-438e2897f1b9"
|
||||
version = "0.5.1"
|
||||
|
||||
[[deps.ChainRules]]
|
||||
deps = ["Adapt", "ChainRulesCore", "Compat", "Distributed", "GPUArraysCore", "IrrationalConstants", "LinearAlgebra", "Random", "RealDot", "SparseArrays", "Statistics", "StructArrays"]
|
||||
git-tree-sha1 = "1cdf290d4feec68824bfb84f4bfc9f3aba185647"
|
||||
uuid = "082447d4-558c-5d27-93f4-14fc19e9eca2"
|
||||
version = "1.51.1"
|
||||
|
||||
[[deps.ChainRulesCore]]
|
||||
deps = ["Compat", "LinearAlgebra", "SparseArrays"]
|
||||
git-tree-sha1 = "e30f2f4e20f7f186dc36529910beaedc60cfa644"
|
||||
uuid = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
|
||||
version = "1.16.0"
|
||||
|
||||
[[deps.CommonSubexpressions]]
|
||||
deps = ["MacroTools", "Test"]
|
||||
git-tree-sha1 = "7b8a93dba8af7e3b42fecabf646260105ac373f7"
|
||||
uuid = "bbf7d656-a473-5ed7-a52c-81e309532950"
|
||||
version = "0.3.0"
|
||||
|
||||
[[deps.Compat]]
|
||||
deps = ["UUIDs"]
|
||||
git-tree-sha1 = "4e88377ae7ebeaf29a047aa1ee40826e0b708a5d"
|
||||
uuid = "34da2185-b29b-5c13-b0c7-acf172513d20"
|
||||
version = "4.7.0"
|
||||
weakdeps = ["Dates", "LinearAlgebra"]
|
||||
|
||||
[deps.Compat.extensions]
|
||||
CompatLinearAlgebraExt = "LinearAlgebra"
|
||||
|
||||
[[deps.CompilerSupportLibraries_jll]]
|
||||
deps = ["Artifacts", "Libdl"]
|
||||
uuid = "e66e0078-7015-5450-92f7-15fbd957f2ae"
|
||||
version = "1.0.5+0"
|
||||
|
||||
[[deps.CompositionsBase]]
|
||||
git-tree-sha1 = "802bb88cd69dfd1509f6670416bd4434015693ad"
|
||||
uuid = "a33af91c-f02d-484b-be07-31d278c5ca2b"
|
||||
version = "0.1.2"
|
||||
|
||||
[deps.CompositionsBase.extensions]
|
||||
CompositionsBaseInverseFunctionsExt = "InverseFunctions"
|
||||
|
||||
[deps.CompositionsBase.weakdeps]
|
||||
InverseFunctions = "3587e190-3f89-42d0-90ee-14403ec27112"
|
||||
|
||||
[[deps.ConstructionBase]]
|
||||
deps = ["LinearAlgebra"]
|
||||
git-tree-sha1 = "738fec4d684a9a6ee9598a8bfee305b26831f28c"
|
||||
uuid = "187b0558-2788-49d3-abe0-74a17ed4e7c9"
|
||||
version = "1.5.2"
|
||||
|
||||
[deps.ConstructionBase.extensions]
|
||||
ConstructionBaseIntervalSetsExt = "IntervalSets"
|
||||
ConstructionBaseStaticArraysExt = "StaticArrays"
|
||||
|
||||
[deps.ConstructionBase.weakdeps]
|
||||
IntervalSets = "8197267c-284f-5f27-9208-e0e47529a953"
|
||||
StaticArrays = "90137ffa-7385-5640-81b9-e52037218182"
|
||||
|
||||
[[deps.ContextVariablesX]]
|
||||
deps = ["Compat", "Logging", "UUIDs"]
|
||||
git-tree-sha1 = "25cc3803f1030ab855e383129dcd3dc294e322cc"
|
||||
uuid = "6add18c4-b38d-439d-96f6-d6bc489c04c5"
|
||||
version = "0.1.3"
|
||||
|
||||
[[deps.DataAPI]]
|
||||
git-tree-sha1 = "8da84edb865b0b5b0100c0666a9bc9a0b71c553c"
|
||||
uuid = "9a962f9c-6df0-11e9-0e5d-c546b8b5ee8a"
|
||||
version = "1.15.0"
|
||||
|
||||
[[deps.DataStructures]]
|
||||
deps = ["Compat", "InteractiveUtils", "OrderedCollections"]
|
||||
git-tree-sha1 = "cf25ccb972fec4e4817764d01c82386ae94f77b4"
|
||||
uuid = "864edb3b-99cc-5e75-8d2d-829cb0a9cfe8"
|
||||
version = "0.18.14"
|
||||
|
||||
[[deps.DataValueInterfaces]]
|
||||
git-tree-sha1 = "bfc1187b79289637fa0ef6d4436ebdfe6905cbd6"
|
||||
uuid = "e2d170a0-9d28-54be-80f0-106bbe20a464"
|
||||
version = "1.0.0"
|
||||
|
||||
[[deps.Dates]]
|
||||
deps = ["Printf"]
|
||||
uuid = "ade2ca70-3891-5945-98fb-dc099432e06a"
|
||||
|
||||
[[deps.DefineSingletons]]
|
||||
git-tree-sha1 = "0fba8b706d0178b4dc7fd44a96a92382c9065c2c"
|
||||
uuid = "244e2a9f-e319-4986-a169-4d1fe445cd52"
|
||||
version = "0.1.2"
|
||||
|
||||
[[deps.DelimitedFiles]]
|
||||
deps = ["Mmap"]
|
||||
git-tree-sha1 = "9e2f36d3c96a820c678f2f1f1782582fcf685bae"
|
||||
uuid = "8bb1440f-4735-579b-a4ab-409b98df4dab"
|
||||
version = "1.9.1"
|
||||
|
||||
[[deps.DiffResults]]
|
||||
deps = ["StaticArraysCore"]
|
||||
git-tree-sha1 = "782dd5f4561f5d267313f23853baaaa4c52ea621"
|
||||
uuid = "163ba53b-c6d8-5494-b064-1a9d43ac40c5"
|
||||
version = "1.1.0"
|
||||
|
||||
[[deps.DiffRules]]
|
||||
deps = ["IrrationalConstants", "LogExpFunctions", "NaNMath", "Random", "SpecialFunctions"]
|
||||
git-tree-sha1 = "23163d55f885173722d1e4cf0f6110cdbaf7e272"
|
||||
uuid = "b552c78f-8df3-52c6-915a-8e097449b14b"
|
||||
version = "1.15.1"
|
||||
|
||||
[[deps.Distributed]]
|
||||
deps = ["Random", "Serialization", "Sockets"]
|
||||
uuid = "8ba89e20-285c-5b6f-9357-94700520ee1b"
|
||||
|
||||
[[deps.Distributions]]
|
||||
deps = ["FillArrays", "LinearAlgebra", "PDMats", "Printf", "QuadGK", "Random", "SparseArrays", "SpecialFunctions", "Statistics", "StatsAPI", "StatsBase", "StatsFuns", "Test"]
|
||||
git-tree-sha1 = "e76a3281de2719d7c81ed62c6ea7057380c87b1d"
|
||||
uuid = "31c24e10-a181-5473-b8eb-7969acd0382f"
|
||||
version = "0.25.98"
|
||||
|
||||
[deps.Distributions.extensions]
|
||||
DistributionsChainRulesCoreExt = "ChainRulesCore"
|
||||
DistributionsDensityInterfaceExt = "DensityInterface"
|
||||
|
||||
[deps.Distributions.weakdeps]
|
||||
ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
|
||||
DensityInterface = "b429d917-457f-4dbc-8f4c-0cc954292b1d"
|
||||
|
||||
[[deps.DocStringExtensions]]
|
||||
deps = ["LibGit2"]
|
||||
git-tree-sha1 = "2fb1e02f2b635d0845df5d7c167fec4dd739b00d"
|
||||
uuid = "ffbed154-4ef7-542d-bbb7-c09d3a79fcae"
|
||||
version = "0.9.3"
|
||||
|
||||
[[deps.Downloads]]
|
||||
deps = ["ArgTools", "FileWatching", "LibCURL", "NetworkOptions"]
|
||||
uuid = "f43a241f-c20a-4ad4-852c-f6b1247861c6"
|
||||
version = "1.6.0"
|
||||
|
||||
[[deps.DualNumbers]]
|
||||
deps = ["Calculus", "NaNMath", "SpecialFunctions"]
|
||||
git-tree-sha1 = "5837a837389fccf076445fce071c8ddaea35a566"
|
||||
uuid = "fa6b7ba4-c1ee-5f82-b5fc-ecf0adba8f74"
|
||||
version = "0.6.8"
|
||||
|
||||
[[deps.ExprTools]]
|
||||
git-tree-sha1 = "c1d06d129da9f55715c6c212866f5b1bddc5fa00"
|
||||
uuid = "e2ba6199-217a-4e67-a87a-7c52f15ade04"
|
||||
version = "0.1.9"
|
||||
|
||||
[[deps.FLoops]]
|
||||
deps = ["BangBang", "Compat", "FLoopsBase", "InitialValues", "JuliaVariables", "MLStyle", "Serialization", "Setfield", "Transducers"]
|
||||
git-tree-sha1 = "ffb97765602e3cbe59a0589d237bf07f245a8576"
|
||||
uuid = "cc61a311-1640-44b5-9fba-1b764f453329"
|
||||
version = "0.2.1"
|
||||
|
||||
[[deps.FLoopsBase]]
|
||||
deps = ["ContextVariablesX"]
|
||||
git-tree-sha1 = "656f7a6859be8673bf1f35da5670246b923964f7"
|
||||
uuid = "b9860ae5-e623-471e-878b-f6a53c775ea6"
|
||||
version = "0.1.1"
|
||||
|
||||
[[deps.FileWatching]]
|
||||
uuid = "7b1f6079-737a-58dc-b8bc-7a2ca5c1b5ee"
|
||||
|
||||
[[deps.FillArrays]]
|
||||
deps = ["LinearAlgebra", "Random", "SparseArrays", "Statistics"]
|
||||
git-tree-sha1 = "0b3b52afd0f87b0a3f5ada0466352d125c9db458"
|
||||
uuid = "1a297f60-69ca-5386-bcde-b61e274b549b"
|
||||
version = "1.2.1"
|
||||
|
||||
[[deps.Flux]]
|
||||
deps = ["Adapt", "CUDA", "ChainRulesCore", "Functors", "LinearAlgebra", "MLUtils", "MacroTools", "NNlib", "NNlibCUDA", "OneHotArrays", "Optimisers", "Preferences", "ProgressLogging", "Random", "Reexport", "SparseArrays", "SpecialFunctions", "Statistics", "Zygote", "cuDNN"]
|
||||
git-tree-sha1 = "3e2c3704c2173ab4b1935362384ca878b53d4c34"
|
||||
uuid = "587475ba-b771-5e3f-ad9e-33799f191a9c"
|
||||
version = "0.13.17"
|
||||
|
||||
[deps.Flux.extensions]
|
||||
AMDGPUExt = "AMDGPU"
|
||||
FluxMetalExt = "Metal"
|
||||
|
||||
[deps.Flux.weakdeps]
|
||||
AMDGPU = "21141c5a-9bdb-4563-92ae-f87d6854732e"
|
||||
Metal = "dde4c033-4e86-420c-a63e-0dd931031962"
|
||||
|
||||
[[deps.ForwardDiff]]
|
||||
deps = ["CommonSubexpressions", "DiffResults", "DiffRules", "LinearAlgebra", "LogExpFunctions", "NaNMath", "Preferences", "Printf", "Random", "SpecialFunctions"]
|
||||
git-tree-sha1 = "00e252f4d706b3d55a8863432e742bf5717b498d"
|
||||
uuid = "f6369f11-7733-5829-9624-2563aa707210"
|
||||
version = "0.10.35"
|
||||
weakdeps = ["StaticArrays"]
|
||||
|
||||
[deps.ForwardDiff.extensions]
|
||||
ForwardDiffStaticArraysExt = "StaticArrays"
|
||||
|
||||
[[deps.Functors]]
|
||||
deps = ["LinearAlgebra"]
|
||||
git-tree-sha1 = "478f8c3145bb91d82c2cf20433e8c1b30df454cc"
|
||||
uuid = "d9f16b24-f501-4c13-a1f2-28368ffc5196"
|
||||
version = "0.4.4"
|
||||
|
||||
[[deps.Future]]
|
||||
deps = ["Random"]
|
||||
uuid = "9fa8497b-333b-5362-9e8d-4d0656e87820"
|
||||
|
||||
[[deps.GPUArrays]]
|
||||
deps = ["Adapt", "GPUArraysCore", "LLVM", "LinearAlgebra", "Printf", "Random", "Reexport", "Serialization", "Statistics"]
|
||||
git-tree-sha1 = "2e57b4a4f9cc15e85a24d603256fe08e527f48d1"
|
||||
uuid = "0c68f7d7-f131-5f86-a1c3-88cf8149b2d7"
|
||||
version = "8.8.1"
|
||||
|
||||
[[deps.GPUArraysCore]]
|
||||
deps = ["Adapt"]
|
||||
git-tree-sha1 = "2d6ca471a6c7b536127afccfa7564b5b39227fe0"
|
||||
uuid = "46192b85-c4d5-4398-a991-12ede77f4527"
|
||||
version = "0.1.5"
|
||||
|
||||
[[deps.GPUCompiler]]
|
||||
deps = ["ExprTools", "InteractiveUtils", "LLVM", "Libdl", "Logging", "Scratch", "TimerOutputs", "UUIDs"]
|
||||
git-tree-sha1 = "cb090aea21c6ca78d59672a7e7d13bd56d09de64"
|
||||
uuid = "61eb1bfa-7361-4325-ad38-22787b887f55"
|
||||
version = "0.20.3"
|
||||
|
||||
[[deps.GeneralUtils]]
|
||||
deps = ["CUDA", "DataStructures", "Distributions", "Flux", "JSON3"]
|
||||
path = "C:\\Users\\naraw\\.julia\\dev\\GeneralUtils"
|
||||
uuid = "c6c72f09-b708-4ac8-ac7c-2084d70108fe"
|
||||
version = "0.1.0"
|
||||
|
||||
[[deps.HypergeometricFunctions]]
|
||||
deps = ["DualNumbers", "LinearAlgebra", "OpenLibm_jll", "SpecialFunctions"]
|
||||
git-tree-sha1 = "ce7ea9cc5db29563b1fe20196b6d23ab3b111384"
|
||||
uuid = "34004b35-14d8-5ef3-9330-4cdb6864b03a"
|
||||
version = "0.3.18"
|
||||
|
||||
[[deps.IRTools]]
|
||||
deps = ["InteractiveUtils", "MacroTools", "Test"]
|
||||
git-tree-sha1 = "eac00994ce3229a464c2847e956d77a2c64ad3a5"
|
||||
uuid = "7869d1d1-7146-5819-86e3-90919afe41df"
|
||||
version = "0.4.10"
|
||||
|
||||
[[deps.InitialValues]]
|
||||
git-tree-sha1 = "4da0f88e9a39111c2fa3add390ab15f3a44f3ca3"
|
||||
uuid = "22cec73e-a1b8-11e9-2c92-598750a2cf9c"
|
||||
version = "0.3.1"
|
||||
|
||||
[[deps.InteractiveUtils]]
|
||||
deps = ["Markdown"]
|
||||
uuid = "b77e0a4c-d291-57a0-90e8-8db25a27a240"
|
||||
|
||||
[[deps.IrrationalConstants]]
|
||||
git-tree-sha1 = "630b497eafcc20001bba38a4651b327dcfc491d2"
|
||||
uuid = "92d709cd-6900-40b7-9082-c6be49f344b6"
|
||||
version = "0.2.2"
|
||||
|
||||
[[deps.IteratorInterfaceExtensions]]
|
||||
git-tree-sha1 = "a3f24677c21f5bbe9d2a714f95dcd58337fb2856"
|
||||
uuid = "82899510-4779-5014-852e-03e436cf321d"
|
||||
version = "1.0.0"
|
||||
|
||||
[[deps.JLLWrappers]]
|
||||
deps = ["Preferences"]
|
||||
git-tree-sha1 = "abc9885a7ca2052a736a600f7fa66209f96506e1"
|
||||
uuid = "692b3bcd-3c85-4b1f-b108-f13ce0eb3210"
|
||||
version = "1.4.1"
|
||||
|
||||
[[deps.JSON3]]
|
||||
deps = ["Dates", "Mmap", "Parsers", "PrecompileTools", "StructTypes", "UUIDs"]
|
||||
git-tree-sha1 = "5b62d93f2582b09e469b3099d839c2d2ebf5066d"
|
||||
uuid = "0f8b85d8-7281-11e9-16c2-39a750bddbf1"
|
||||
version = "1.13.1"
|
||||
|
||||
[[deps.JuliaVariables]]
|
||||
deps = ["MLStyle", "NameResolution"]
|
||||
git-tree-sha1 = "49fb3cb53362ddadb4415e9b73926d6b40709e70"
|
||||
uuid = "b14d175d-62b4-44ba-8fb7-3064adc8c3ec"
|
||||
version = "0.2.4"
|
||||
|
||||
[[deps.KernelAbstractions]]
|
||||
deps = ["Adapt", "Atomix", "InteractiveUtils", "LinearAlgebra", "MacroTools", "PrecompileTools", "SparseArrays", "StaticArrays", "UUIDs", "UnsafeAtomics", "UnsafeAtomicsLLVM"]
|
||||
git-tree-sha1 = "b48617c5d764908b5fac493cd907cf33cc11eec1"
|
||||
uuid = "63c18a36-062a-441e-b654-da1e3ab1ce7c"
|
||||
version = "0.9.6"
|
||||
|
||||
[[deps.LLVM]]
|
||||
deps = ["CEnum", "LLVMExtra_jll", "Libdl", "Printf", "Unicode"]
|
||||
git-tree-sha1 = "5007c1421563108110bbd57f63d8ad4565808818"
|
||||
uuid = "929cbde3-209d-540e-8aea-75f648917ca0"
|
||||
version = "5.2.0"
|
||||
|
||||
[[deps.LLVMExtra_jll]]
|
||||
deps = ["Artifacts", "JLLWrappers", "LazyArtifacts", "Libdl", "TOML"]
|
||||
git-tree-sha1 = "1222116d7313cdefecf3d45a2bc1a89c4e7c9217"
|
||||
uuid = "dad2f222-ce93-54a1-a47d-0025e8a3acab"
|
||||
version = "0.0.22+0"
|
||||
|
||||
[[deps.LazyArtifacts]]
|
||||
deps = ["Artifacts", "Pkg"]
|
||||
uuid = "4af54fe1-eca0-43a8-85a7-787d91b784e3"
|
||||
|
||||
[[deps.LibCURL]]
|
||||
deps = ["LibCURL_jll", "MozillaCACerts_jll"]
|
||||
uuid = "b27032c2-a3e7-50c8-80cd-2d36dbcbfd21"
|
||||
version = "0.6.3"
|
||||
|
||||
[[deps.LibCURL_jll]]
|
||||
deps = ["Artifacts", "LibSSH2_jll", "Libdl", "MbedTLS_jll", "Zlib_jll", "nghttp2_jll"]
|
||||
uuid = "deac9b47-8bc7-5906-a0fe-35ac56dc84c0"
|
||||
version = "7.84.0+0"
|
||||
|
||||
[[deps.LibGit2]]
|
||||
deps = ["Base64", "NetworkOptions", "Printf", "SHA"]
|
||||
uuid = "76f85450-5226-5b5a-8eaa-529ad045b433"
|
||||
|
||||
[[deps.LibSSH2_jll]]
|
||||
deps = ["Artifacts", "Libdl", "MbedTLS_jll"]
|
||||
uuid = "29816b5a-b9ab-546f-933c-edad1886dfa8"
|
||||
version = "1.10.2+0"
|
||||
|
||||
[[deps.Libdl]]
|
||||
uuid = "8f399da3-3557-5675-b5ff-fb832c97cbdb"
|
||||
|
||||
[[deps.LinearAlgebra]]
|
||||
deps = ["Libdl", "OpenBLAS_jll", "libblastrampoline_jll"]
|
||||
uuid = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
|
||||
|
||||
[[deps.LogExpFunctions]]
|
||||
deps = ["DocStringExtensions", "IrrationalConstants", "LinearAlgebra"]
|
||||
git-tree-sha1 = "c3ce8e7420b3a6e071e0fe4745f5d4300e37b13f"
|
||||
uuid = "2ab3a3ac-af41-5b50-aa03-7779005ae688"
|
||||
version = "0.3.24"
|
||||
|
||||
[deps.LogExpFunctions.extensions]
|
||||
LogExpFunctionsChainRulesCoreExt = "ChainRulesCore"
|
||||
LogExpFunctionsChangesOfVariablesExt = "ChangesOfVariables"
|
||||
LogExpFunctionsInverseFunctionsExt = "InverseFunctions"
|
||||
|
||||
[deps.LogExpFunctions.weakdeps]
|
||||
ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
|
||||
ChangesOfVariables = "9e997f8a-9a97-42d5-a9f1-ce6bfc15e2c0"
|
||||
InverseFunctions = "3587e190-3f89-42d0-90ee-14403ec27112"
|
||||
|
||||
[[deps.Logging]]
|
||||
uuid = "56ddb016-857b-54e1-b83d-db4d58db5568"
|
||||
|
||||
[[deps.MLStyle]]
|
||||
git-tree-sha1 = "bc38dff0548128765760c79eb7388a4b37fae2c8"
|
||||
uuid = "d8e11817-5142-5d16-987a-aa16d5891078"
|
||||
version = "0.4.17"
|
||||
|
||||
[[deps.MLUtils]]
|
||||
deps = ["ChainRulesCore", "Compat", "DataAPI", "DelimitedFiles", "FLoops", "NNlib", "Random", "ShowCases", "SimpleTraits", "Statistics", "StatsBase", "Tables", "Transducers"]
|
||||
git-tree-sha1 = "3504cdb8c2bc05bde4d4b09a81b01df88fcbbba0"
|
||||
uuid = "f1d291b0-491e-4a28-83b9-f70985020b54"
|
||||
version = "0.4.3"
|
||||
|
||||
[[deps.MacroTools]]
|
||||
deps = ["Markdown", "Random"]
|
||||
git-tree-sha1 = "42324d08725e200c23d4dfb549e0d5d89dede2d2"
|
||||
uuid = "1914dd2f-81c6-5fcd-8719-6d5c9610ff09"
|
||||
version = "0.5.10"
|
||||
|
||||
[[deps.Markdown]]
|
||||
deps = ["Base64"]
|
||||
uuid = "d6f4376e-aef5-505a-96c1-9c027394607a"
|
||||
|
||||
[[deps.MbedTLS_jll]]
|
||||
deps = ["Artifacts", "Libdl"]
|
||||
uuid = "c8ffd9c3-330d-5841-b78e-0817d7145fa1"
|
||||
version = "2.28.2+0"
|
||||
|
||||
[[deps.MicroCollections]]
|
||||
deps = ["BangBang", "InitialValues", "Setfield"]
|
||||
git-tree-sha1 = "629afd7d10dbc6935ec59b32daeb33bc4460a42e"
|
||||
uuid = "128add7d-3638-4c79-886c-908ea0c25c34"
|
||||
version = "0.1.4"
|
||||
|
||||
[[deps.Missings]]
|
||||
deps = ["DataAPI"]
|
||||
git-tree-sha1 = "f66bdc5de519e8f8ae43bdc598782d35a25b1272"
|
||||
uuid = "e1d29d7a-bbdc-5cf2-9ac0-f12de2c33e28"
|
||||
version = "1.1.0"
|
||||
|
||||
[[deps.Mmap]]
|
||||
uuid = "a63ad114-7e13-5084-954f-fe012c677804"
|
||||
|
||||
[[deps.MozillaCACerts_jll]]
|
||||
uuid = "14a3606d-f60d-562e-9121-12d972cd8159"
|
||||
version = "2022.10.11"
|
||||
|
||||
[[deps.NNlib]]
|
||||
deps = ["Adapt", "Atomix", "ChainRulesCore", "GPUArraysCore", "KernelAbstractions", "LinearAlgebra", "Pkg", "Random", "Requires", "Statistics"]
|
||||
git-tree-sha1 = "72240e3f5ca031937bd536182cb2c031da5f46dd"
|
||||
uuid = "872c559c-99b0-510c-b3b7-b6c96a88d5cd"
|
||||
version = "0.8.21"
|
||||
|
||||
[deps.NNlib.extensions]
|
||||
NNlibAMDGPUExt = "AMDGPU"
|
||||
|
||||
[deps.NNlib.weakdeps]
|
||||
AMDGPU = "21141c5a-9bdb-4563-92ae-f87d6854732e"
|
||||
|
||||
[[deps.NNlibCUDA]]
|
||||
deps = ["Adapt", "CUDA", "LinearAlgebra", "NNlib", "Random", "Statistics", "cuDNN"]
|
||||
git-tree-sha1 = "f94a9684394ff0d325cc12b06da7032d8be01aaf"
|
||||
uuid = "a00861dc-f156-4864-bf3c-e6376f28a68d"
|
||||
version = "0.2.7"
|
||||
|
||||
[[deps.NaNMath]]
|
||||
deps = ["OpenLibm_jll"]
|
||||
git-tree-sha1 = "0877504529a3e5c3343c6f8b4c0381e57e4387e4"
|
||||
uuid = "77ba4419-2d1f-58cd-9bb1-8ffee604a2e3"
|
||||
version = "1.0.2"
|
||||
|
||||
[[deps.NameResolution]]
|
||||
deps = ["PrettyPrint"]
|
||||
git-tree-sha1 = "1a0fa0e9613f46c9b8c11eee38ebb4f590013c5e"
|
||||
uuid = "71a1bf82-56d0-4bbc-8a3c-48b961074391"
|
||||
version = "0.1.5"
|
||||
|
||||
[[deps.NetworkOptions]]
|
||||
uuid = "ca575930-c2e3-43a9-ace4-1e988b2c1908"
|
||||
version = "1.2.0"
|
||||
|
||||
[[deps.OneHotArrays]]
|
||||
deps = ["Adapt", "ChainRulesCore", "Compat", "GPUArraysCore", "LinearAlgebra", "NNlib"]
|
||||
git-tree-sha1 = "5e4029759e8699ec12ebdf8721e51a659443403c"
|
||||
uuid = "0b1bfda6-eb8a-41d2-88d8-f5af5cad476f"
|
||||
version = "0.2.4"
|
||||
|
||||
[[deps.OpenBLAS_jll]]
|
||||
deps = ["Artifacts", "CompilerSupportLibraries_jll", "Libdl"]
|
||||
uuid = "4536629a-c528-5b80-bd46-f80d51c5b363"
|
||||
version = "0.3.21+4"
|
||||
|
||||
[[deps.OpenLibm_jll]]
|
||||
deps = ["Artifacts", "Libdl"]
|
||||
uuid = "05823500-19ac-5b8b-9628-191a04bc5112"
|
||||
version = "0.8.1+0"
|
||||
|
||||
[[deps.OpenSpecFun_jll]]
|
||||
deps = ["Artifacts", "CompilerSupportLibraries_jll", "JLLWrappers", "Libdl", "Pkg"]
|
||||
git-tree-sha1 = "13652491f6856acfd2db29360e1bbcd4565d04f1"
|
||||
uuid = "efe28fd5-8261-553b-a9e1-b2916fc3738e"
|
||||
version = "0.5.5+0"
|
||||
|
||||
[[deps.Optimisers]]
|
||||
deps = ["ChainRulesCore", "Functors", "LinearAlgebra", "Random", "Statistics"]
|
||||
git-tree-sha1 = "6a01f65dd8583dee82eecc2a19b0ff21521aa749"
|
||||
uuid = "3bd65402-5787-11e9-1adc-39752487f4e2"
|
||||
version = "0.2.18"
|
||||
|
||||
[[deps.OrderedCollections]]
|
||||
git-tree-sha1 = "d321bf2de576bf25ec4d3e4360faca399afca282"
|
||||
uuid = "bac558e1-5e72-5ebc-8fee-abe8a469f55d"
|
||||
version = "1.6.0"
|
||||
|
||||
[[deps.PDMats]]
|
||||
deps = ["LinearAlgebra", "SparseArrays", "SuiteSparse"]
|
||||
git-tree-sha1 = "67eae2738d63117a196f497d7db789821bce61d1"
|
||||
uuid = "90014a1f-27ba-587c-ab20-58faa44d9150"
|
||||
version = "0.11.17"
|
||||
|
||||
[[deps.Parsers]]
|
||||
deps = ["Dates", "PrecompileTools", "UUIDs"]
|
||||
git-tree-sha1 = "4b2e829ee66d4218e0cef22c0a64ee37cf258c29"
|
||||
uuid = "69de0a69-1ddd-5017-9359-2bf0b02dc9f0"
|
||||
version = "2.7.1"
|
||||
|
||||
[[deps.Pkg]]
|
||||
deps = ["Artifacts", "Dates", "Downloads", "FileWatching", "LibGit2", "Libdl", "Logging", "Markdown", "Printf", "REPL", "Random", "SHA", "Serialization", "TOML", "Tar", "UUIDs", "p7zip_jll"]
|
||||
uuid = "44cfe95a-1eb2-52ea-b672-e2afdf69b78f"
|
||||
version = "1.9.2"
|
||||
|
||||
[[deps.PrecompileTools]]
|
||||
deps = ["Preferences"]
|
||||
git-tree-sha1 = "9673d39decc5feece56ef3940e5dafba15ba0f81"
|
||||
uuid = "aea7be01-6a6a-4083-8856-8a6e6704d82a"
|
||||
version = "1.1.2"
|
||||
|
||||
[[deps.Preferences]]
|
||||
deps = ["TOML"]
|
||||
git-tree-sha1 = "7eb1686b4f04b82f96ed7a4ea5890a4f0c7a09f1"
|
||||
uuid = "21216c6a-2e73-6563-6e65-726566657250"
|
||||
version = "1.4.0"
|
||||
|
||||
[[deps.PrettyPrint]]
|
||||
git-tree-sha1 = "632eb4abab3449ab30c5e1afaa874f0b98b586e4"
|
||||
uuid = "8162dcfd-2161-5ef2-ae6c-7681170c5f98"
|
||||
version = "0.2.0"
|
||||
|
||||
[[deps.Printf]]
|
||||
deps = ["Unicode"]
|
||||
uuid = "de0858da-6303-5e67-8744-51eddeeeb8d7"
|
||||
|
||||
[[deps.ProgressLogging]]
|
||||
deps = ["Logging", "SHA", "UUIDs"]
|
||||
git-tree-sha1 = "80d919dee55b9c50e8d9e2da5eeafff3fe58b539"
|
||||
uuid = "33c8b6b6-d38a-422a-b730-caa89a2f386c"
|
||||
version = "0.1.4"
|
||||
|
||||
[[deps.QuadGK]]
|
||||
deps = ["DataStructures", "LinearAlgebra"]
|
||||
git-tree-sha1 = "6ec7ac8412e83d57e313393220879ede1740f9ee"
|
||||
uuid = "1fd47b50-473d-5c70-9696-f719f8f3bcdc"
|
||||
version = "2.8.2"
|
||||
|
||||
[[deps.REPL]]
|
||||
deps = ["InteractiveUtils", "Markdown", "Sockets", "Unicode"]
|
||||
uuid = "3fa0cd96-eef1-5676-8a61-b3b8758bbffb"
|
||||
|
||||
[[deps.Random]]
|
||||
deps = ["SHA", "Serialization"]
|
||||
uuid = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
|
||||
|
||||
[[deps.Random123]]
|
||||
deps = ["Random", "RandomNumbers"]
|
||||
git-tree-sha1 = "552f30e847641591ba3f39fd1bed559b9deb0ef3"
|
||||
uuid = "74087812-796a-5b5d-8853-05524746bad3"
|
||||
version = "1.6.1"
|
||||
|
||||
[[deps.RandomNumbers]]
|
||||
deps = ["Random", "Requires"]
|
||||
git-tree-sha1 = "043da614cc7e95c703498a491e2c21f58a2b8111"
|
||||
uuid = "e6cf234a-135c-5ec9-84dd-332b85af5143"
|
||||
version = "1.5.3"
|
||||
|
||||
[[deps.RealDot]]
|
||||
deps = ["LinearAlgebra"]
|
||||
git-tree-sha1 = "9f0a1b71baaf7650f4fa8a1d168c7fb6ee41f0c9"
|
||||
uuid = "c1ae055f-0cd5-4b69-90a6-9a35b1a98df9"
|
||||
version = "0.1.0"
|
||||
|
||||
[[deps.Reexport]]
|
||||
git-tree-sha1 = "45e428421666073eab6f2da5c9d310d99bb12f9b"
|
||||
uuid = "189a3867-3050-52da-a836-e630ba90ab69"
|
||||
version = "1.2.2"
|
||||
|
||||
[[deps.Requires]]
|
||||
deps = ["UUIDs"]
|
||||
git-tree-sha1 = "838a3a4188e2ded87a4f9f184b4b0d78a1e91cb7"
|
||||
uuid = "ae029012-a4dd-5104-9daa-d747884805df"
|
||||
version = "1.3.0"
|
||||
|
||||
[[deps.Rmath]]
|
||||
deps = ["Random", "Rmath_jll"]
|
||||
git-tree-sha1 = "f65dcb5fa46aee0cf9ed6274ccbd597adc49aa7b"
|
||||
uuid = "79098fc4-a85e-5d69-aa6a-4863f24498fa"
|
||||
version = "0.7.1"
|
||||
|
||||
[[deps.Rmath_jll]]
|
||||
deps = ["Artifacts", "JLLWrappers", "Libdl", "Pkg"]
|
||||
git-tree-sha1 = "6ed52fdd3382cf21947b15e8870ac0ddbff736da"
|
||||
uuid = "f50d1b31-88e8-58de-be2c-1cc44531875f"
|
||||
version = "0.4.0+0"
|
||||
|
||||
[[deps.SHA]]
|
||||
uuid = "ea8e919c-243c-51af-8825-aaa63cd721ce"
|
||||
version = "0.7.0"
|
||||
|
||||
[[deps.Scratch]]
|
||||
deps = ["Dates"]
|
||||
git-tree-sha1 = "30449ee12237627992a99d5e30ae63e4d78cd24a"
|
||||
uuid = "6c6a2e73-6563-6170-7368-637461726353"
|
||||
version = "1.2.0"
|
||||
|
||||
[[deps.Serialization]]
|
||||
uuid = "9e88b42a-f829-5b0c-bbe9-9e923198166b"
|
||||
|
||||
[[deps.Setfield]]
|
||||
deps = ["ConstructionBase", "Future", "MacroTools", "StaticArraysCore"]
|
||||
git-tree-sha1 = "e2cc6d8c88613c05e1defb55170bf5ff211fbeac"
|
||||
uuid = "efcf1570-3423-57d1-acb7-fd33fddbac46"
|
||||
version = "1.1.1"
|
||||
|
||||
[[deps.ShowCases]]
|
||||
git-tree-sha1 = "7f534ad62ab2bd48591bdeac81994ea8c445e4a5"
|
||||
uuid = "605ecd9f-84a6-4c9e-81e2-4798472b76a3"
|
||||
version = "0.1.0"
|
||||
|
||||
[[deps.SimpleTraits]]
|
||||
deps = ["InteractiveUtils", "MacroTools"]
|
||||
git-tree-sha1 = "5d7e3f4e11935503d3ecaf7186eac40602e7d231"
|
||||
uuid = "699a6c99-e7fa-54fc-8d76-47d257e15c1d"
|
||||
version = "0.9.4"
|
||||
|
||||
[[deps.Sockets]]
|
||||
uuid = "6462fe0b-24de-5631-8697-dd941f90decc"
|
||||
|
||||
[[deps.SortingAlgorithms]]
|
||||
deps = ["DataStructures"]
|
||||
git-tree-sha1 = "c60ec5c62180f27efea3ba2908480f8055e17cee"
|
||||
uuid = "a2af1166-a08f-5f64-846c-94a0d3cef48c"
|
||||
version = "1.1.1"
|
||||
|
||||
[[deps.SparseArrays]]
|
||||
deps = ["Libdl", "LinearAlgebra", "Random", "Serialization", "SuiteSparse_jll"]
|
||||
uuid = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
|
||||
|
||||
[[deps.SpecialFunctions]]
|
||||
deps = ["IrrationalConstants", "LogExpFunctions", "OpenLibm_jll", "OpenSpecFun_jll"]
|
||||
git-tree-sha1 = "7beb031cf8145577fbccacd94b8a8f4ce78428d3"
|
||||
uuid = "276daf66-3868-5448-9aa4-cd146d93841b"
|
||||
version = "2.3.0"
|
||||
weakdeps = ["ChainRulesCore"]
|
||||
|
||||
[deps.SpecialFunctions.extensions]
|
||||
SpecialFunctionsChainRulesCoreExt = "ChainRulesCore"
|
||||
|
||||
[[deps.SplittablesBase]]
|
||||
deps = ["Setfield", "Test"]
|
||||
git-tree-sha1 = "e08a62abc517eb79667d0a29dc08a3b589516bb5"
|
||||
uuid = "171d559e-b47b-412a-8079-5efa626c420e"
|
||||
version = "0.1.15"
|
||||
|
||||
[[deps.StaticArrays]]
|
||||
deps = ["LinearAlgebra", "Random", "StaticArraysCore", "Statistics"]
|
||||
git-tree-sha1 = "832afbae2a45b4ae7e831f86965469a24d1d8a83"
|
||||
uuid = "90137ffa-7385-5640-81b9-e52037218182"
|
||||
version = "1.5.26"
|
||||
|
||||
[[deps.StaticArraysCore]]
|
||||
git-tree-sha1 = "6b7ba252635a5eff6a0b0664a41ee140a1c9e72a"
|
||||
uuid = "1e83bf80-4336-4d27-bf5d-d5a4f845583c"
|
||||
version = "1.4.0"
|
||||
|
||||
[[deps.Statistics]]
|
||||
deps = ["LinearAlgebra", "SparseArrays"]
|
||||
uuid = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
|
||||
version = "1.9.0"
|
||||
|
||||
[[deps.StatsAPI]]
|
||||
deps = ["LinearAlgebra"]
|
||||
git-tree-sha1 = "45a7769a04a3cf80da1c1c7c60caf932e6f4c9f7"
|
||||
uuid = "82ae8749-77ed-4fe6-ae5f-f523153014b0"
|
||||
version = "1.6.0"
|
||||
|
||||
[[deps.StatsBase]]
|
||||
deps = ["DataAPI", "DataStructures", "LinearAlgebra", "LogExpFunctions", "Missings", "Printf", "Random", "SortingAlgorithms", "SparseArrays", "Statistics", "StatsAPI"]
|
||||
git-tree-sha1 = "75ebe04c5bed70b91614d684259b661c9e6274a4"
|
||||
uuid = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
|
||||
version = "0.34.0"
|
||||
|
||||
[[deps.StatsFuns]]
|
||||
deps = ["HypergeometricFunctions", "IrrationalConstants", "LogExpFunctions", "Reexport", "Rmath", "SpecialFunctions"]
|
||||
git-tree-sha1 = "f625d686d5a88bcd2b15cd81f18f98186fdc0c9a"
|
||||
uuid = "4c63d2b9-4356-54db-8cca-17b64c39e42c"
|
||||
version = "1.3.0"
|
||||
|
||||
[deps.StatsFuns.extensions]
|
||||
StatsFunsChainRulesCoreExt = "ChainRulesCore"
|
||||
StatsFunsInverseFunctionsExt = "InverseFunctions"
|
||||
|
||||
[deps.StatsFuns.weakdeps]
|
||||
ChainRulesCore = "d360d2e6-b24c-11e9-a2a3-2a2ae2dbcce4"
|
||||
InverseFunctions = "3587e190-3f89-42d0-90ee-14403ec27112"
|
||||
|
||||
[[deps.StructArrays]]
|
||||
deps = ["Adapt", "DataAPI", "GPUArraysCore", "StaticArraysCore", "Tables"]
|
||||
git-tree-sha1 = "521a0e828e98bb69042fec1809c1b5a680eb7389"
|
||||
uuid = "09ab397b-f2b6-538f-b94a-2f83cf4a842a"
|
||||
version = "0.6.15"
|
||||
|
||||
[[deps.StructTypes]]
|
||||
deps = ["Dates", "UUIDs"]
|
||||
git-tree-sha1 = "ca4bccb03acf9faaf4137a9abc1881ed1841aa70"
|
||||
uuid = "856f2bd8-1eba-4b0a-8007-ebc267875bd4"
|
||||
version = "1.10.0"
|
||||
|
||||
[[deps.SuiteSparse]]
|
||||
deps = ["Libdl", "LinearAlgebra", "Serialization", "SparseArrays"]
|
||||
uuid = "4607b0f0-06f3-5cda-b6b1-a6196a1729e9"
|
||||
|
||||
[[deps.SuiteSparse_jll]]
|
||||
deps = ["Artifacts", "Libdl", "Pkg", "libblastrampoline_jll"]
|
||||
uuid = "bea87d4a-7f5b-5778-9afe-8cc45184846c"
|
||||
version = "5.10.1+6"
|
||||
|
||||
[[deps.TOML]]
|
||||
deps = ["Dates"]
|
||||
uuid = "fa267f1f-6049-4f14-aa54-33bafae1ed76"
|
||||
version = "1.0.3"
|
||||
|
||||
[[deps.TableTraits]]
|
||||
deps = ["IteratorInterfaceExtensions"]
|
||||
git-tree-sha1 = "c06b2f539df1c6efa794486abfb6ed2022561a39"
|
||||
uuid = "3783bdb8-4a98-5b6b-af9a-565f29a5fe9c"
|
||||
version = "1.0.1"
|
||||
|
||||
[[deps.Tables]]
|
||||
deps = ["DataAPI", "DataValueInterfaces", "IteratorInterfaceExtensions", "LinearAlgebra", "OrderedCollections", "TableTraits", "Test"]
|
||||
git-tree-sha1 = "1544b926975372da01227b382066ab70e574a3ec"
|
||||
uuid = "bd369af6-aec1-5ad0-b16a-f7cc5008161c"
|
||||
version = "1.10.1"
|
||||
|
||||
[[deps.Tar]]
|
||||
deps = ["ArgTools", "SHA"]
|
||||
uuid = "a4e569a6-e804-4fa4-b0f3-eef7a1d5b13e"
|
||||
version = "1.10.0"
|
||||
|
||||
[[deps.Test]]
|
||||
deps = ["InteractiveUtils", "Logging", "Random", "Serialization"]
|
||||
uuid = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
|
||||
|
||||
[[deps.TimerOutputs]]
|
||||
deps = ["ExprTools", "Printf"]
|
||||
git-tree-sha1 = "f548a9e9c490030e545f72074a41edfd0e5bcdd7"
|
||||
uuid = "a759f4b9-e2f1-59dc-863e-4aeb61b1ea8f"
|
||||
version = "0.5.23"
|
||||
|
||||
[[deps.Transducers]]
|
||||
deps = ["Adapt", "ArgCheck", "BangBang", "Baselet", "CompositionsBase", "DefineSingletons", "Distributed", "InitialValues", "Logging", "Markdown", "MicroCollections", "Requires", "Setfield", "SplittablesBase", "Tables"]
|
||||
git-tree-sha1 = "a66fb81baec325cf6ccafa243af573b031e87b00"
|
||||
uuid = "28d57a85-8fef-5791-bfe6-a80928e7c999"
|
||||
version = "0.4.77"
|
||||
|
||||
[deps.Transducers.extensions]
|
||||
TransducersBlockArraysExt = "BlockArrays"
|
||||
TransducersDataFramesExt = "DataFrames"
|
||||
TransducersLazyArraysExt = "LazyArrays"
|
||||
TransducersOnlineStatsBaseExt = "OnlineStatsBase"
|
||||
TransducersReferenceablesExt = "Referenceables"
|
||||
|
||||
[deps.Transducers.weakdeps]
|
||||
BlockArrays = "8e7c35d0-a365-5155-bbbb-fb81a777f24e"
|
||||
DataFrames = "a93c6f00-e57d-5684-b7b6-d8193f3e46c0"
|
||||
LazyArrays = "5078a376-72f3-5289-bfd5-ec5146d43c02"
|
||||
OnlineStatsBase = "925886fa-5bf2-5e8e-b522-a9147a512338"
|
||||
Referenceables = "42d2dcc6-99eb-4e98-b66c-637b7d73030e"
|
||||
|
||||
[[deps.UUIDs]]
|
||||
deps = ["Random", "SHA"]
|
||||
uuid = "cf7118a7-6976-5b1a-9a39-7adc72f591a4"
|
||||
|
||||
[[deps.Unicode]]
|
||||
uuid = "4ec0a83e-493e-50e2-b9ac-8f72acf5a8f5"
|
||||
|
||||
[[deps.UnsafeAtomics]]
|
||||
git-tree-sha1 = "6331ac3440856ea1988316b46045303bef658278"
|
||||
uuid = "013be700-e6cd-48c3-b4a1-df204f14c38f"
|
||||
version = "0.2.1"
|
||||
|
||||
[[deps.UnsafeAtomicsLLVM]]
|
||||
deps = ["LLVM", "UnsafeAtomics"]
|
||||
git-tree-sha1 = "ea37e6066bf194ab78f4e747f5245261f17a7175"
|
||||
uuid = "d80eeb9a-aca5-4d75-85e5-170c8b632249"
|
||||
version = "0.1.2"
|
||||
|
||||
[[deps.Zlib_jll]]
|
||||
deps = ["Libdl"]
|
||||
uuid = "83775a58-1f1d-513f-b197-d71354ab007a"
|
||||
version = "1.2.13+0"
|
||||
|
||||
[[deps.Zygote]]
|
||||
deps = ["AbstractFFTs", "ChainRules", "ChainRulesCore", "DiffRules", "Distributed", "FillArrays", "ForwardDiff", "GPUArrays", "GPUArraysCore", "IRTools", "InteractiveUtils", "LinearAlgebra", "LogExpFunctions", "MacroTools", "NaNMath", "PrecompileTools", "Random", "Requires", "SparseArrays", "SpecialFunctions", "Statistics", "ZygoteRules"]
|
||||
git-tree-sha1 = "5be3ddb88fc992a7d8ea96c3f10a49a7e98ebc7b"
|
||||
uuid = "e88e6eb3-aa80-5325-afca-941959d7151f"
|
||||
version = "0.6.62"
|
||||
|
||||
[deps.Zygote.extensions]
|
||||
ZygoteColorsExt = "Colors"
|
||||
ZygoteDistancesExt = "Distances"
|
||||
ZygoteTrackerExt = "Tracker"
|
||||
|
||||
[deps.Zygote.weakdeps]
|
||||
Colors = "5ae59095-9a9b-59fe-a467-6f913c188581"
|
||||
Distances = "b4f34e82-e78d-54a5-968a-f98e89d6e8f7"
|
||||
Tracker = "9f7883ad-71c0-57eb-9f7f-b5c9e6d3789c"
|
||||
|
||||
[[deps.ZygoteRules]]
|
||||
deps = ["ChainRulesCore", "MacroTools"]
|
||||
git-tree-sha1 = "977aed5d006b840e2e40c0b48984f7463109046d"
|
||||
uuid = "700de1a5-db45-46bc-99cf-38207098b444"
|
||||
version = "0.2.3"
|
||||
|
||||
[[deps.cuDNN]]
|
||||
deps = ["CEnum", "CUDA", "CUDNN_jll"]
|
||||
git-tree-sha1 = "f65490d187861d6222cb38bcbbff3fd949a7ec3e"
|
||||
uuid = "02a925ec-e4fe-4b08-9a7e-0d78e3d38ccd"
|
||||
version = "1.0.4"
|
||||
|
||||
[[deps.libblastrampoline_jll]]
|
||||
deps = ["Artifacts", "Libdl"]
|
||||
uuid = "8e850b90-86db-534c-a0d3-1478176c7d93"
|
||||
version = "5.8.0+0"
|
||||
|
||||
[[deps.nghttp2_jll]]
|
||||
deps = ["Artifacts", "Libdl"]
|
||||
uuid = "8e850ede-7688-5339-a07c-302acd2aaf8d"
|
||||
version = "1.48.0+0"
|
||||
|
||||
[[deps.p7zip_jll]]
|
||||
deps = ["Artifacts", "Libdl"]
|
||||
uuid = "3f19e933-33d8-53b3-aaab-bd5110c3b7a0"
|
||||
version = "17.4.0+0"
|
||||
14
previousVersion/0.0.05-alpha/Project.toml
Normal file
14
previousVersion/0.0.05-alpha/Project.toml
Normal file
@@ -0,0 +1,14 @@
|
||||
name = "IronpenGPU"
|
||||
uuid = "3d5396ea-818e-43fc-a9d3-164248e840cd"
|
||||
authors = ["ton <narawat@gmail.com>"]
|
||||
version = "0.1.0"
|
||||
|
||||
[deps]
|
||||
CUDA = "052768ef-5323-5732-b1bb-66c8b64840ba"
|
||||
Dates = "ade2ca70-3891-5945-98fb-dc099432e06a"
|
||||
Flux = "587475ba-b771-5e3f-ad9e-33799f191a9c"
|
||||
GeneralUtils = "c6c72f09-b708-4ac8-ac7c-2084d70108fe"
|
||||
JSON3 = "0f8b85d8-7281-11e9-16c2-39a750bddbf1"
|
||||
LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
|
||||
Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
|
||||
Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
|
||||
839
previousVersion/0.0.05-alpha/main_gpu_1.jl
Normal file
839
previousVersion/0.0.05-alpha/main_gpu_1.jl
Normal file
@@ -0,0 +1,839 @@
|
||||
using Pkg; Pkg.activate("."); Pkg.resolve(), Pkg.instantiate()
|
||||
using Revise
|
||||
using BenchmarkTools, Cthulhu
|
||||
using Flux, CUDA
|
||||
using BSON, JSON3
|
||||
using MLDatasets: MNIST
|
||||
using MLUtils, ProgressMeter, Dates, Random,
|
||||
Serialization, OneHotArrays , GLMakie
|
||||
|
||||
using GeneralUtils
|
||||
using CondaPkg, PythonCall
|
||||
|
||||
|
||||
# # if one need to reinstall all python packages
|
||||
# channels = ["anaconda", "conda-forge", "pytorch"]
|
||||
# for i in channels CondaPkg.add_channel(i) end
|
||||
# condapackage = ["numpy", "pytorch", "snntorch"]
|
||||
# for i in condapackage try CondaPkg.rm(i) catch end end
|
||||
# for i in condapackage CondaPkg.add(i) end
|
||||
|
||||
np = pyimport("numpy")
|
||||
torch = pyimport("torch")
|
||||
spikegen = pyimport("snntorch.spikegen") # https://github.com/jeshraghian/snntorch
|
||||
|
||||
using IronpenGPU
|
||||
using GeneralUtils
|
||||
|
||||
sep = Sys.iswindows() ? "\\" : "/"
|
||||
rootDir = pwd()
|
||||
|
||||
# select compute device
|
||||
# device = Flux.CUDA.functional() ? gpu : cpu # Flux provide "cpu" and "gpu" keywork
|
||||
device = gpu
|
||||
if device == gpu CUDA.device!(1) end #CHANGE
|
||||
# CUDA.allowscalar(false) # turn off scalar indexing in CPU to make it easier when moving to GPU
|
||||
#------------------------------------------------------------------------------------------------100
|
||||
|
||||
|
||||
|
||||
"""
|
||||
Todo:
|
||||
- []
|
||||
|
||||
Change from version:
|
||||
-
|
||||
|
||||
All features
|
||||
-
|
||||
"""
|
||||
|
||||
|
||||
# communication config --------------------------------------------------------------------------100
|
||||
|
||||
database_ip = "localhost"
|
||||
# database_ip = "192.168.0.8"
|
||||
|
||||
#------------------------------------------------------------------------------------------------100
|
||||
modelname = "runOn_gpu_1" #CHANGE
|
||||
imageBatch = 1
|
||||
|
||||
|
||||
function generate_snn(filename::String, location::String)
|
||||
signalInput_portnumbers = (10, 20, imageBatch) # 2nd dim needs to match
|
||||
# input signal + copied input signal + noise.
|
||||
# 3rd dim is input batch size
|
||||
noise_portnumbers = (signalInput_portnumbers[1], 1)
|
||||
output_portnumbers = (10, 1)
|
||||
|
||||
# 5000 neurons are maximum for 64GB memory i.e. 300 LIF : 200 ALIF
|
||||
lif_neuron_number = (signalInput_portnumbers[1], 3) # CHANGE
|
||||
alif_neuron_number = (signalInput_portnumbers[1], 2) # CHANGE from Allen Institute, ALIF is 20-40% of LIF
|
||||
|
||||
# totalNeurons = computeNeuronNumber + noise_portnumbers + signalInput_portnumbers
|
||||
# totalInputPort = noise_portnumbers + signalInput_portnumbers
|
||||
|
||||
# kfn and neuron config
|
||||
passthrough_neuron_params = Dict(
|
||||
:type => "passthroughNeuron"
|
||||
)
|
||||
|
||||
lif_neuron_params = Dict{Symbol, Any}(
|
||||
:type => "lifNeuron",
|
||||
:v_t_default => 0.0,
|
||||
:v_th => 1.0, # neuron firing threshold (this value is treated as maximum bound if I use auto generate)
|
||||
:tau_m => 50.0, # membrane time constant in millisecond.
|
||||
:eta => 1e-6,
|
||||
# Good starting value is 1/10th of tau_a
|
||||
# This is problem specific parameter. It controls how leaky the neuron is.
|
||||
# Too high(less leaky) makes learning algo harder to move model into direction that reduce error
|
||||
# resulting in model's error to explode exponantially likely because learning algo will try to
|
||||
# exert more force (larger w_out_change) to move neuron into direction that reduce error
|
||||
# For example, model error from 7 to 2e6.
|
||||
|
||||
:synapticConnectionPercent => 50, # % coverage of total neurons in kfn
|
||||
)
|
||||
|
||||
alif_neuron_params = Dict{Symbol, Any}(
|
||||
:type => "alifNeuron",
|
||||
:v_t_default => 0.0,
|
||||
:v_th => 1.0, # neuron firing threshold (this value is treated as maximum bound if I use auto generate)
|
||||
:tau_m => 50.0, # membrane time constant in millisecond.
|
||||
:eta => 1e-6,
|
||||
# Good starting value is 1/10th of tau_a
|
||||
# This is problem specific parameter. It controls how leaky the neuron is.
|
||||
# Too high(less leaky) makes learning algo harder to move model into direction that reduce error
|
||||
# resulting in model's error to explode exponantially likely because learning algo will try to
|
||||
# exert more force (larger w_out_change) to move neuron into direction that reduce error
|
||||
# For example, model error from 7 to 2e6.
|
||||
|
||||
:tau_a => 800.0, # adaptation time constant in millisecond. it defines neuron memory length.
|
||||
# This is problem specific parameter
|
||||
# Good starting value is 0.5 to 2 times of info STORE-RECALL length i.e. total time SNN takes to
|
||||
# perform a task, for example, equals to episode length.
|
||||
# From "Spike frequency adaptation supports network computations on temporally dispersed
|
||||
# information"
|
||||
|
||||
:synapticConnectionPercent => 50, # % coverage of total neurons in kfn
|
||||
)
|
||||
|
||||
linear_neuron_params = Dict{Symbol, Any}(
|
||||
:type => "linearNeuron",
|
||||
:v_th => 1.0, # neuron firing threshold (this value is treated as maximum bound if I use auto generate)
|
||||
:tau_out => 100.0, # output time constant in millisecond.
|
||||
:synapticConnectionPercent => 50, # % coverage of total neurons in kfn
|
||||
# Good starting value is 1/50th of tau_a
|
||||
# This is problem specific parameter.
|
||||
# It controls how leaky the neuron is.
|
||||
# Too high(less leaky) makes learning algo harder to move model into direction that reduce error
|
||||
# resulting in model's error to explode exponantially. For example, model error from 7 to 2e6
|
||||
# One can image training output neuron is like Tetris Game.
|
||||
)
|
||||
|
||||
integrate_neuron_params = Dict{Symbol, Any}(
|
||||
:type => "integrateNeuron",
|
||||
:synapticConnectionPercent => 100, # % coverage of total neurons in kfn
|
||||
:eta => 1e-6,
|
||||
:tau_out => 100.0,
|
||||
# Good starting value is 1/50th of tau_a
|
||||
# This is problem specific parameter.
|
||||
# It controls how leaky the neuron is.
|
||||
# Too high(less leaky) makes learning algo harder to move model into direction that reduce error
|
||||
# resulting in model's error to explode exponantially. For example, model error from 7 to 2e6
|
||||
# One can image training output neuron is like Tetris Game.
|
||||
)
|
||||
|
||||
I_kfnparams = Dict{Symbol, Any}(
|
||||
:knowledgeFnName=> "I",
|
||||
:neuronFiringRateTarget=> 20.0, # Hz
|
||||
|
||||
# group relavent info
|
||||
:inputPort=> Dict(
|
||||
:noise=> Dict(
|
||||
:numbers=> noise_portnumbers,
|
||||
:params=> passthrough_neuron_params,
|
||||
),
|
||||
:signal=> Dict(
|
||||
:numbers=> signalInput_portnumbers, # in case of GloVe word encoding, it is 300
|
||||
:params=> passthrough_neuron_params,
|
||||
),
|
||||
),
|
||||
:outputPort=> Dict(
|
||||
:numbers=> output_portnumbers, # output neuron, this is also the output length
|
||||
:params=> linear_neuron_params,
|
||||
),
|
||||
:computeNeuron=> Dict(
|
||||
:lif=> Dict(
|
||||
:numbers=> lif_neuron_number, # number in (row, col) tuple format
|
||||
:params=> lif_neuron_params,
|
||||
),
|
||||
:alif=> Dict(
|
||||
:numbers=> alif_neuron_number, # number in (row, col) tuple format
|
||||
:params=> alif_neuron_params,
|
||||
),
|
||||
),
|
||||
)
|
||||
|
||||
#------------------------------------------------------------------------------------------------100
|
||||
|
||||
model = IronpenGPU.kfn_1(I_kfnparams, device=device);
|
||||
|
||||
|
||||
# serialize(location * sep * filename, model)
|
||||
println("SNN generated")
|
||||
|
||||
return model
|
||||
end
|
||||
|
||||
function data_loader()
|
||||
# test problem
|
||||
trainDataset = MNIST(:train)[1:3] # total 60000
|
||||
# validateDataset = MNIST(:test)
|
||||
validateDataset = MNIST(:train)[1:3]
|
||||
labelDict = [0:9...]
|
||||
|
||||
trainData = MLUtils.DataLoader(
|
||||
trainDataset; # fullTrainDataset or trainDataset
|
||||
batchsize=imageBatch,
|
||||
collate=true,
|
||||
shuffle=true,
|
||||
buffer=true,
|
||||
partial=false, # better for gpu memory if batchsize is fixed
|
||||
# parallel=true, #BUG ?? causing dataloader into forever loop
|
||||
)
|
||||
|
||||
validateData = MLUtils.DataLoader(
|
||||
validateDataset;
|
||||
batchsize=imageBatch,
|
||||
collate=true,
|
||||
shuffle=true,
|
||||
buffer=true,
|
||||
partial=false, # better for gpu memory if batchsize is fixed
|
||||
# parallel=true, #BUG ?? causing dataloader into forever loop
|
||||
)
|
||||
|
||||
# dummy data used to debug
|
||||
# trainData = [(rand(10, 10), [5]), (rand(10, 10), [2])]
|
||||
# trainData = [(rand(10, 10), [5]),]
|
||||
|
||||
return trainData, validateData, labelDict
|
||||
end
|
||||
|
||||
function train_snn(model, trainData, validateData, labelDict::Vector)
|
||||
|
||||
# random seed
|
||||
# rng = MersenneTwister(1234)
|
||||
|
||||
logitLog = zeros(10, 2)
|
||||
firedNeurons_t1 = zeros(1)
|
||||
var1 = zeros(10, 2)
|
||||
var2 = zeros(10, 2)
|
||||
var3 = zeros(10, 2)
|
||||
var4 = zeros(10, 2)
|
||||
|
||||
# ----------------------------------- plot ----------------------------------- #
|
||||
plot10 = Observable(firedNeurons_t1)
|
||||
|
||||
plot20 = Observable(logitLog[1 , :])
|
||||
plot21 = Observable(logitLog[2 , :])
|
||||
plot22 = Observable(logitLog[3 , :])
|
||||
plot23 = Observable(logitLog[4 , :])
|
||||
plot24 = Observable(logitLog[5 , :])
|
||||
plot25 = Observable(logitLog[6 , :])
|
||||
plot26 = Observable(logitLog[7 , :])
|
||||
plot27 = Observable(logitLog[8 , :])
|
||||
plot28 = Observable(logitLog[9 , :])
|
||||
plot29 = Observable(logitLog[10, :])
|
||||
|
||||
plot30 = Observable(var1[1 , :])
|
||||
plot31 = Observable(var1[2 , :])
|
||||
plot32 = Observable(var1[3 , :])
|
||||
plot33 = Observable(var1[4 , :])
|
||||
plot34 = Observable(var1[5 , :])
|
||||
plot35 = Observable(var1[6 , :])
|
||||
plot36 = Observable(var1[7 , :])
|
||||
plot37 = Observable(var1[8 , :])
|
||||
plot38 = Observable(var1[9 , :])
|
||||
plot39 = Observable(var1[10, :])
|
||||
|
||||
plot40 = Observable(var2[1 , :])
|
||||
plot41 = Observable(var2[2 , :])
|
||||
plot42 = Observable(var2[3 , :])
|
||||
plot43 = Observable(var2[4 , :])
|
||||
plot44 = Observable(var2[5 , :])
|
||||
plot45 = Observable(var2[6 , :])
|
||||
plot46 = Observable(var2[7 , :])
|
||||
plot47 = Observable(var2[8 , :])
|
||||
plot48 = Observable(var2[9 , :])
|
||||
plot49 = Observable(var2[10, :])
|
||||
|
||||
plot50 = Observable(var3[1 , :])
|
||||
plot51 = Observable(var3[2 , :])
|
||||
plot52 = Observable(var3[3 , :])
|
||||
plot53 = Observable(var3[4 , :])
|
||||
plot54 = Observable(var3[5 , :])
|
||||
plot55 = Observable(var3[6 , :])
|
||||
plot56 = Observable(var3[7 , :])
|
||||
plot57 = Observable(var3[8 , :])
|
||||
plot58 = Observable(var3[9 , :])
|
||||
plot59 = Observable(var3[10, :])
|
||||
|
||||
plot60 = Observable(var4[1 , :])
|
||||
plot61 = Observable(var4[2 , :])
|
||||
plot62 = Observable(var4[3 , :])
|
||||
plot63 = Observable(var4[4 , :])
|
||||
plot64 = Observable(var4[5 , :])
|
||||
plot65 = Observable(var4[6 , :])
|
||||
plot66 = Observable(var4[7 , :])
|
||||
plot67 = Observable(var4[8 , :])
|
||||
plot68 = Observable(var4[9 , :])
|
||||
plot69 = Observable(var4[10, :])
|
||||
|
||||
# main figure
|
||||
fig1 = Figure()
|
||||
|
||||
subfig1 = GLMakie.Axis(fig1[1, 1], # define position of this subfigure inside a figure
|
||||
title = "RSNN firedNeurons_t1",
|
||||
xlabel = "time",
|
||||
ylabel = "data"
|
||||
)
|
||||
lines!(subfig1, plot10, label = "firedNeurons_t1")
|
||||
# axislegend(subfig1, position = :lb)
|
||||
|
||||
subfig2 = GLMakie.Axis(fig1[2, 1], # define position of this subfigure inside a figure
|
||||
title = "output neurons logit",
|
||||
xlabel = "time",
|
||||
ylabel = "data"
|
||||
)
|
||||
|
||||
lines!(subfig2, plot20, label = "0", color = 1, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig2, plot21, label = "1", color = 2, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig2, plot22, label = "2", color = 3, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig2, plot23, label = "3", color = 4, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig2, plot24, label = "4", color = 5, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig2, plot25, label = "5", color = 6, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig2, plot26, label = "6", color = 7, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig2, plot27, label = "7", color = 8, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig2, plot28, label = "8", color = 9, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig2, plot29, label = "9", color = 10, colormap = :tab10, colorrange = (1, 10))
|
||||
# axislegend(subfig2, position = :lb)
|
||||
|
||||
|
||||
subfig3 = GLMakie.Axis(fig1[3, 1], # define position of this subfigure inside a figure
|
||||
title = "last RSNN wRec",
|
||||
xlabel = "time",
|
||||
ylabel = "data"
|
||||
)
|
||||
lines!(subfig3, plot30, label = "0", color = 1, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig3, plot31, label = "1", color = 2, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig3, plot32, label = "2", color = 3, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig3, plot33, label = "3", color = 4, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig3, plot34, label = "4", color = 5, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig3, plot35, label = "5", color = 6, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig3, plot36, label = "6", color = 7, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig3, plot37, label = "7", color = 8, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig3, plot38, label = "8", color = 9, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig3, plot39, label = "9", color = 10, colormap = :tab10, colorrange = (1, 10))
|
||||
# axislegend(subfig3, position = :lb)
|
||||
|
||||
subfig4 = GLMakie.Axis(fig1[4, 1], # define position of this subfigure inside a figure
|
||||
title = "RSNN v_t1",
|
||||
xlabel = "time",
|
||||
ylabel = "data"
|
||||
)
|
||||
lines!(subfig4, plot40, label = "0", color = 1, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig4, plot41, label = "1", color = 2, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig4, plot42, label = "2", color = 3, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig4, plot43, label = "3", color = 4, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig4, plot44, label = "4", color = 5, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig4, plot45, label = "5", color = 6, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig4, plot46, label = "6", color = 7, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig4, plot47, label = "7", color = 8, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig4, plot48, label = "8", color = 9, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig4, plot49, label = "9", color = 10, colormap = :tab10, colorrange = (1, 10))
|
||||
# axislegend(subfig4, position = :lb)
|
||||
|
||||
subfig5 = GLMakie.Axis(fig1[5, 1], # define position of this subfigure inside a figure
|
||||
title = "output neuron epsilonRec",
|
||||
xlabel = "time",
|
||||
ylabel = "data"
|
||||
)
|
||||
lines!(subfig5, plot50, label = "0", color = 1, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig5, plot51, label = "1", color = 2, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig5, plot52, label = "2", color = 3, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig5, plot53, label = "3", color = 4, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig5, plot54, label = "4", color = 5, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig5, plot55, label = "5", color = 6, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig5, plot56, label = "6", color = 7, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig5, plot57, label = "7", color = 8, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig5, plot58, label = "8", color = 9, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig5, plot59, label = "9", color = 10, colormap = :tab10, colorrange = (1, 10))
|
||||
# axislegend(subfig5, position = :lb)
|
||||
|
||||
subfig6 = GLMakie.Axis(fig1[6, 1], # define position of this subfigure inside a figure
|
||||
title = "output neuron wRecChange",
|
||||
xlabel = "time",
|
||||
ylabel = "data"
|
||||
)
|
||||
lines!(subfig6, plot60, label = "0", color = 1, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig6, plot61, label = "1", color = 2, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig6, plot62, label = "2", color = 3, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig6, plot63, label = "3", color = 4, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig6, plot64, label = "4", color = 5, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig6, plot65, label = "5", color = 6, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig6, plot66, label = "6", color = 7, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig6, plot67, label = "7", color = 8, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig6, plot68, label = "8", color = 9, colormap = :tab10, colorrange = (1, 10) )
|
||||
lines!(subfig6, plot69, label = "9", color = 10, colormap = :tab10, colorrange = (1, 10))
|
||||
# axislegend(subfig6, position = :lb)
|
||||
|
||||
# wait(display(fig1))
|
||||
# display(fig1)
|
||||
# --------------------------------- end plot --------------------------------- #
|
||||
|
||||
# model learning
|
||||
thinkingPeriod = 16 # 1000-784 = 216
|
||||
bestAccuracy = 0.0
|
||||
finalAnswer = [0] |> device # store model prediction in (logit of choices, batch)
|
||||
stop = 0
|
||||
for epoch = 1:1000
|
||||
stop == 3 ? break : false
|
||||
println("epoch $epoch")
|
||||
n = length(trainData)
|
||||
println("n $n")
|
||||
p = Progress(n, dt=1.0) # minimum update interval: 1 second
|
||||
for (imgBatch, labels) in trainData # imgBatch (28, 28, 4) i.e. (row, col, batch)
|
||||
stop == 3 ? break : false
|
||||
# signal (10, 2, 784, 4) i.e. (row, col, timestep, batch)
|
||||
signal = spikeGenerator(imgBatch, [0.05, 0.1, 0.2, 0.3, 0.5], noise=(true, 1, 1.0), copies=18)
|
||||
if length(size(signal)) == 3
|
||||
row, col, sequence = size(signal)
|
||||
batch = 1
|
||||
else
|
||||
row, col, sequence, batch = size(signal)
|
||||
end
|
||||
|
||||
# encode labels
|
||||
correctAnswer = onehotbatch(labels, labelDict) # (choices, batch)
|
||||
|
||||
# insert data into model sequencially
|
||||
for timestep in 1:(sequence + thinkingPeriod) # sMNIST has 784 timestep(pixel) + thinking period = 1000 timestep
|
||||
if timestep <= sequence
|
||||
current_pixel = view(signal, :, :, timestep, :) |> device
|
||||
else
|
||||
current_pixel = zeros(row, col, batch) |> device # dummy input in "thinking" period
|
||||
end
|
||||
|
||||
if timestep == 1 # tell a model to start learning. 1-time only
|
||||
model.learningStage = [1]
|
||||
finalAnswer = [0] |> device
|
||||
elseif timestep == (sequence+thinkingPeriod)
|
||||
model.learningStage = [3]
|
||||
else
|
||||
end
|
||||
|
||||
# predict
|
||||
logit, _firedNeurons_t1 = model(current_pixel)
|
||||
|
||||
# # log answer of all timestep
|
||||
# logitLog = [logitLog;; logit]
|
||||
# firedNeurons_t1 = push!(firedNeurons_t1, _firedNeurons_t1)
|
||||
# var1 = [var1;; _var1]
|
||||
# var2 = [var2;; _var2]
|
||||
# var3 = [var3;; _var3]
|
||||
# var4 = [var4;; _var4]
|
||||
|
||||
if timestep < sequence # online learning, 1-by-1 timestep
|
||||
# no error calculation
|
||||
elseif timestep == sequence # online learning, 1-by-1 timestep
|
||||
# no error calculation
|
||||
elseif timestep > sequence && timestep < sequence+thinkingPeriod # collect answer
|
||||
finalAnswer = length(finalAnswer) == 1 ? logit : finalAnswer .+ logit # (logit, batch)
|
||||
predict_cpu = logit |> cpu
|
||||
|
||||
# predict = mapslices(GeneralUtils.vectorMax, on_zt_cpu; dims=1)
|
||||
modelError = (predict_cpu .- correctAnswer) |> device
|
||||
outputError = (predict_cpu .- correctAnswer) |> device
|
||||
|
||||
lif_epsilonRec_cpu = model.lif_epsilonRec |> cpu
|
||||
on_zt_cpu = model.on_zt |> cpu
|
||||
|
||||
IronpenGPU.compute_paramsChange!(model, modelError, outputError)
|
||||
|
||||
lif_wRecChange_cpu = model.lif_wRecChange |> cpu
|
||||
|
||||
|
||||
# if sum(lif_wRecChange_cpu) != 0
|
||||
# println("")
|
||||
# lif_vt_cpu = model.lif_vt |> cpu
|
||||
# lif_zt_cpu = model.lif_zt |> cpu
|
||||
|
||||
# lif_recSignal = model.lif_recSignal |> cpu
|
||||
|
||||
# on_vt_cpu = model.on_vt |> cpu
|
||||
# on_vt_cpu = on_vt_cpu[1,1,:,1]
|
||||
|
||||
# on_zt_cpu = on_zt_cpu[1,1,:,1]
|
||||
# on_wOutChange_cpu = model.on_wOutChange |> cpu
|
||||
# on_wOutChange_cpu = sum(on_wOutChange_cpu, dims=(1,2))
|
||||
# println("lif vt $(lif_vt_cpu[1,1,5,1]) lif zt $(lif_zt_cpu[1,1,5,1]) on_vt $on_vt_cpu on_zt $on_zt_cpu on_wOutChange_cpu $on_wOutChange_cpu")
|
||||
# println("lif_recSignal ", lif_recSignal)
|
||||
# println("")
|
||||
# println("lif_epsilonRec_cpu ", lif_epsilonRec_cpu)
|
||||
# println("")
|
||||
# println("lif_wRecChange ", lif_wRecChange_cpu)
|
||||
# println("")
|
||||
# zit_cumulative = model.zit_cumulative |> cpu
|
||||
# println("zit_cumulative ", zit_cumulative)
|
||||
|
||||
# # error("DEBUG -> main $(Dates.now())")
|
||||
# end
|
||||
|
||||
elseif timestep == sequence+thinkingPeriod
|
||||
finalAnswer = length(finalAnswer) == 1 ? logit : finalAnswer .+ logit # (logit, batch)
|
||||
predict_cpu = logit |> cpu
|
||||
|
||||
# predict = mapslices(GeneralUtils.vectorMax, on_zt_cpu; dims=1)
|
||||
modelError = (predict_cpu .- correctAnswer) |> device
|
||||
outputError = (predict_cpu .- correctAnswer) |> device
|
||||
|
||||
lif_epsilonRec_cpu = model.lif_epsilonRec |> cpu
|
||||
on_zt_cpu = model.on_zt |> cpu
|
||||
|
||||
IronpenGPU.compute_paramsChange!(model, modelError, outputError)
|
||||
|
||||
lif_wRecChange_cpu = model.lif_wRecChange |> cpu
|
||||
|
||||
|
||||
println("")
|
||||
lif_recSignal_cpu = model.lif_recSignal |> cpu
|
||||
lif_recSignal_cpu = sum(lif_recSignal_cpu[:,:,5,1])
|
||||
lif_vt_cpu = model.lif_vt |> cpu
|
||||
lif_vt_cpu = lif_vt_cpu[1,1,5,1]
|
||||
lif_zt_cpu = model.lif_zt |> cpu
|
||||
lif_zt_cpu = lif_zt_cpu[1,1,5,1]
|
||||
lif_epsilonRec_cpu = model.lif_epsilonRec |> cpu
|
||||
lif_epsilonRec_cpu = sum(lif_epsilonRec_cpu[:,:,5,1])
|
||||
lif_wRecChange_cpu = sum(lif_wRecChange_cpu[:,:,5,1])
|
||||
on_vt_cpu = model.on_vt |> cpu
|
||||
on_vt_cpu = on_vt_cpu[1,1,:,1]
|
||||
|
||||
on_zt_cpu = on_zt_cpu[1,1,:,1]
|
||||
on_wOutChange_cpu = model.on_wOutChange |> cpu
|
||||
on_wOutChange_cpu = sum(on_wOutChange_cpu, dims=(1,2))
|
||||
println("lif recSignal $lif_recSignal_cpu lif vt $lif_vt_cpu lif zt $lif_zt_cpu lif_epsilonRec_cpu $lif_epsilonRec_cpu lif_wRecChange_cpu $lif_wRecChange_cpu on_vt $on_vt_cpu on_zt $on_zt_cpu on_wOutChange_cpu $on_wOutChange_cpu")
|
||||
# println("lif_recSignal ", lif_recSignal)
|
||||
# println("")
|
||||
# println("lif_epsilonRec_cpu ", lif_epsilonRec_cpu)
|
||||
# println("")
|
||||
# println("lif_wRecChange ", lif_wRecChange_cpu)
|
||||
# println("")
|
||||
# zit_cumulative = model.zit_cumulative |> cpu
|
||||
# println("zit_cumulative ", zit_cumulative)
|
||||
|
||||
# error("DEBUG -> main $(Dates.now())")
|
||||
|
||||
# commit learned weight only if the model answer incorrectly
|
||||
finalAnswer_cpu = finalAnswer |> cpu
|
||||
println("label $(labels[1]) finalAnswer $finalAnswer_cpu")
|
||||
max = isequal.(finalAnswer_cpu[:,1], maximum(finalAnswer_cpu[:,1]))
|
||||
if sum(max) == 1 && findall(max)[1] -1 == labels[1]
|
||||
finalAnswer_cpu = findall(max)[1] - 1
|
||||
println("OK")
|
||||
# println("label $(labels[1]) finalAnswer $finalAnswer_cpu")
|
||||
|
||||
elseif sum(max) == 1 && findall(max)[1] -1 != labels[1]
|
||||
finalAnswer = findall(max)[1] - 1
|
||||
IronpenGPU.learn!(model)
|
||||
println("LEARNING")
|
||||
# println("label $(labels[1]) finalAnswer $finalAnswer_cpu LEARNING")
|
||||
else
|
||||
IronpenGPU.learn!(model)
|
||||
println("LEARNING")
|
||||
# println("epoch $epoch label $(labels[1]) finalAnswer $finalAnswer_cpu LEARNING")
|
||||
end
|
||||
|
||||
# error("DEBUG -> main $(Dates.now())")
|
||||
else
|
||||
error("undefined condition line $(@__LINE__)")
|
||||
# error("DEBUG -> main $(Dates.now())")
|
||||
end
|
||||
|
||||
# update plot
|
||||
# plot10[] = firedNeurons_t1
|
||||
|
||||
# plot20[] = view(logitLog, 1 , :)
|
||||
# plot21[] = view(logitLog, 2 , :)
|
||||
# plot22[] = view(logitLog, 3 , :)
|
||||
# plot23[] = view(logitLog, 4 , :)
|
||||
# plot24[] = view(logitLog, 5 , :)
|
||||
# plot25[] = view(logitLog, 6 , :)
|
||||
# plot26[] = view(logitLog, 7 , :)
|
||||
# plot27[] = view(logitLog, 8 , :)
|
||||
# plot28[] = view(logitLog, 9 , :)
|
||||
# plot29[] = view(logitLog, 10, :)
|
||||
|
||||
# plot30[] = view(var1, 1 , :)
|
||||
# plot31[] = view(var1, 2 , :)
|
||||
# plot32[] = view(var1, 3 , :)
|
||||
# plot33[] = view(var1, 4 , :)
|
||||
# plot34[] = view(var1, 5 , :)
|
||||
# plot35[] = view(var1, 6 , :)
|
||||
# plot36[] = view(var1, 7 , :)
|
||||
# plot37[] = view(var1, 8 , :)
|
||||
# plot38[] = view(var1, 9 , :)
|
||||
# plot39[] = view(var1, 10, :)
|
||||
|
||||
# plot40[] = view(var2, 1 , :)
|
||||
# plot41[] = view(var2, 2 , :)
|
||||
# plot42[] = view(var2, 3 , :)
|
||||
# plot43[] = view(var2, 4 , :)
|
||||
# plot44[] = view(var2, 5 , :)
|
||||
# plot45[] = view(var2, 6 , :)
|
||||
# plot46[] = view(var2, 7 , :)
|
||||
# plot47[] = view(var2, 8 , :)
|
||||
# plot48[] = view(var2, 9 , :)
|
||||
# plot49[] = view(var2, 10, :)
|
||||
|
||||
# plot50[] = view(var3, 1 , :)
|
||||
# plot51[] = view(var3, 2 , :)
|
||||
# plot52[] = view(var3, 3 , :)
|
||||
# plot53[] = view(var3, 4 , :)
|
||||
# plot54[] = view(var3, 5 , :)
|
||||
# plot55[] = view(var3, 6 , :)
|
||||
# plot56[] = view(var3, 7 , :)
|
||||
# plot57[] = view(var3, 8 , :)
|
||||
# plot58[] = view(var3, 9 , :)
|
||||
# plot59[] = view(var3, 10, :)
|
||||
|
||||
# plot60[] = view(var4, 1 , :)
|
||||
# plot61[] = view(var4, 2 , :)
|
||||
# plot62[] = view(var4, 3 , :)
|
||||
# plot63[] = view(var4, 4 , :)
|
||||
# plot64[] = view(var4, 5 , :)
|
||||
# plot65[] = view(var4, 6 , :)
|
||||
# plot66[] = view(var4, 7 , :)
|
||||
# plot67[] = view(var4, 8 , :)
|
||||
# plot68[] = view(var4, 9 , :)
|
||||
# plot69[] = view(var4, 10, :)
|
||||
end
|
||||
# end-thinkingPeriod+2; +2 because initialize logitLog = zeros(10, 2)
|
||||
# _modelRespond = logitLog[:, end-thinkingPeriod+2:end] # answer count during thinking period
|
||||
# _modelRespond = [sum(i) for i in eachrow(_modelRespond)]
|
||||
# modelRespond = isequal.(isequal.(_modelRespond, 0), 0)
|
||||
|
||||
# display(fig1)
|
||||
# sleep(1)
|
||||
# if k % 3 == 0
|
||||
# firedNeurons_t1 = zeros(1)
|
||||
# logitLog = zeros(10, 2)
|
||||
# var1 = zeros(10, 2)
|
||||
# var2 = zeros(10, 2)
|
||||
# var3 = zeros(10, 2)
|
||||
# var4 = zeros(10, 2)
|
||||
# end
|
||||
|
||||
# # if predict == OneHotArrays.onehot(label, labelDict)
|
||||
# # println("model train $label successfully, $k tries")
|
||||
# # # wait(display(fig1))
|
||||
|
||||
# # firedNeurons_t1 = zeros(1)
|
||||
# # logitLog = zeros(10, 2)
|
||||
# # var1 = zeros(10, 2)
|
||||
# # var2 = zeros(10, 2)
|
||||
# # var3 = zeros(10, 2)
|
||||
# # var4 = zeros(10, 2)
|
||||
# # break
|
||||
# # end
|
||||
|
||||
# if k == maxRepeatRound
|
||||
# # println("model train $label unsuccessfully, $maxRepeatRound tries, skip training")
|
||||
# # display(fig1)
|
||||
# firedNeurons_t1 = zeros(1)
|
||||
# logitLog = zeros(10, 2)
|
||||
# var1 = zeros(10, 2)
|
||||
# var2 = zeros(10, 2)
|
||||
# var3 = zeros(10, 2)
|
||||
# var4 = zeros(10, 2)
|
||||
# break
|
||||
# end
|
||||
|
||||
next!(p)
|
||||
end
|
||||
|
||||
if epoch > 200
|
||||
# check accuracy
|
||||
println("validating model")
|
||||
percentCorrect = validate(model, validateData, labelDict)
|
||||
bestAccuracy = percentCorrect > bestAccuracy ? percentCorrect : bestAccuracy
|
||||
println("$modelname model accuracy is $percentCorrect %, best accuracy is $bestAccuracy")
|
||||
end
|
||||
end
|
||||
end
|
||||
|
||||
function validate(model, dataset, labelDict)
|
||||
totalAnswerCorrectly = 0 # score
|
||||
totalSignal = 0
|
||||
thinkingPeriod = 16 # 1000-784 = 216
|
||||
predict = [0] |> device
|
||||
|
||||
n = length(dataset)
|
||||
println("n $n")
|
||||
p = Progress(n, dt=1.0) # minimum update interval: 1 second
|
||||
for (imgBatch, labels) in dataset
|
||||
signal = spikeGenerator(imgBatch, [0.05, 0.1, 0.2, 0.3, 0.5], noise=(true, 1, 0.5), copies=18)
|
||||
if length(size(signal)) == 3
|
||||
row, col, sequence = size(signal)
|
||||
batch = 1
|
||||
else
|
||||
row, col, sequence, batch = size(signal)
|
||||
end
|
||||
|
||||
# encode labels
|
||||
correctAnswer = onehotbatch(labels, labelDict) # (choices, batch)
|
||||
|
||||
# insert data into model sequencially
|
||||
for timestep in 1:(sequence + thinkingPeriod) # sMNIST has 784 timestep(pixel) + thinking period = 1000 timestep
|
||||
if timestep <= sequence
|
||||
current_pixel = view(signal, :, :, timestep, :) |> device
|
||||
else
|
||||
current_pixel = zeros(row, col, batch) |> device # dummy input in "thinking" period
|
||||
end
|
||||
|
||||
if timestep == 1 # tell a model to start learning. 1-time only
|
||||
predict = [0] |> device
|
||||
elseif timestep == (sequence+thinkingPeriod)
|
||||
else
|
||||
end
|
||||
|
||||
# predict
|
||||
logit, _ = model(current_pixel)
|
||||
|
||||
if timestep < sequence # online learning, 1-by-1 timestep
|
||||
# no error calculation
|
||||
elseif timestep == sequence # online learning, 1-by-1 timestep
|
||||
# no error calculation
|
||||
elseif timestep > sequence && timestep < sequence+thinkingPeriod # collect answer
|
||||
predict = length(predict) == 1 ? logit : predict .+ logit # (logit, batch)
|
||||
elseif timestep == sequence+thinkingPeriod
|
||||
predict = length(predict) == 1 ? logit : predict .+ logit # (logit, batch)
|
||||
else
|
||||
error("undefined condition line $(@__LINE__)")
|
||||
end
|
||||
end
|
||||
|
||||
predict_cpu = predict |> cpu
|
||||
_predict_label = mapslices(GeneralUtils.vectorMax, predict_cpu; dims=1)
|
||||
s = sum(_predict_label, dims=1)
|
||||
if 0 ∉ s
|
||||
predict_label = []
|
||||
for i in eachcol(_predict_label)
|
||||
_label = findall(i) .- 1
|
||||
if length(_label) == 1
|
||||
append!(predict_label, _label)
|
||||
else
|
||||
push!(predict_label, -1) # predict more than 1 label. add non-count label.
|
||||
end
|
||||
end
|
||||
answerCorrectly = sum([x == y for (x,y) in zip(predict_label, labels)])
|
||||
totalAnswerCorrectly += answerCorrectly
|
||||
totalSignal += batch
|
||||
end
|
||||
|
||||
next!(p)
|
||||
end
|
||||
|
||||
percentCorrect = totalAnswerCorrectly * 100.0 / totalSignal
|
||||
|
||||
return percentCorrect::Float64
|
||||
end
|
||||
|
||||
""" inputsignals is normal column-major julia matrix in (row, col, batch) dimension
|
||||
- each threshold scan return 2 vectors. 1 for +, 1 for -
|
||||
- noise = (true/false, row, col, probability)
|
||||
"""
|
||||
function spikeGenerator(inputsignals, thresholds=[1.0]; noise=(false, 1, 0.5), copies=0)
|
||||
s = length(size(inputsignals))
|
||||
ar = [] # holding all signals that are scanned
|
||||
for slice in eachslice(inputsignals, dims=s)
|
||||
signal_jl = reshape(slice, (:, 1)) # python array is row-major
|
||||
signal_pytensor = torch.from_numpy( np.asarray(signal_jl) )
|
||||
|
||||
arr = [] # holding signal that is scanned by several thresholds
|
||||
for threshold in thresholds
|
||||
spike_py = spikegen.delta(signal_pytensor, threshold=threshold, off_spike=true)
|
||||
_spike_jl = pyconvert(Array, spike_py.data.numpy())
|
||||
spike_jl = reshape(_spike_jl, (1, :)) # reshape back to julia's column-major
|
||||
spike_jl1 = isequal.(spike_jl, 1)
|
||||
spike_jl2 = isequal.(spike_jl, -1)
|
||||
arr = length(arr) == 0 ? [spike_jl1; spike_jl2] : [arr; spike_jl1; spike_jl2]
|
||||
end
|
||||
arrSize = [size(arr)...]
|
||||
arr = reshape(arr, (arrSize[1], 1, arrSize[2])) # reshape into (row, 1, timestep)
|
||||
|
||||
# multiply col
|
||||
if copies > 0
|
||||
a = deepcopy(arr)
|
||||
for i in 1:copies
|
||||
arr = cat(arr, a, dims=2)
|
||||
end
|
||||
end
|
||||
|
||||
if noise[1] == true
|
||||
arrSize = [size(arr)...]
|
||||
n = noiseGenerator(arrSize[1], noise[2], arrSize[3], prob=noise[3])
|
||||
arr = cat(arr, n, dims=2) # concatenate into (row, signal:noise, timestep)
|
||||
end
|
||||
|
||||
# concatenate into (row, signal:noise, timestep, batch)
|
||||
ar = length(ar) == 0 ? arr : [ar;;;;arr]
|
||||
end
|
||||
return ar
|
||||
end
|
||||
|
||||
function noiseGenerator(row, col, z; prob=0.5)
|
||||
spike_prob = torch.rand(row, col, z) * prob
|
||||
spike_rand = spikegen.rate_conv(spike_prob)
|
||||
noise = isequal.(pyconvert(Array, spike_rand.data.numpy()), 1)
|
||||
|
||||
return noise
|
||||
end
|
||||
|
||||
# function arrayMax(x)
|
||||
# if sum(GeneralUtils.isNotEqual.(x, 0)) == 0 # guard against all-zeros array
|
||||
# return GeneralUtils.isNotEqual.(x, 0)
|
||||
# else
|
||||
# return isequal.(x, maximum(x))
|
||||
# end
|
||||
# end
|
||||
# arraySliceMax(x) = mapslices(arrayMax, x; dims=1)
|
||||
|
||||
function main()
|
||||
filelocation = string(@__DIR__)
|
||||
|
||||
filename = "$modelname.jl163"
|
||||
|
||||
training_start_time = Dates.now()
|
||||
println("$modelname program started $training_start_time")
|
||||
|
||||
model = generate_snn(filename, filelocation)
|
||||
|
||||
trainDataset, validateDataset, labelDict = data_loader()
|
||||
|
||||
train_snn(model, trainDataset, validateDataset, labelDict)
|
||||
|
||||
finish_training_time = Dates.now()
|
||||
println("training done, $training_start_time ==> $finish_training_time ")
|
||||
println(" ///////////////////////////////////////////////////////////////////////")
|
||||
end
|
||||
|
||||
# only runs main() if julia isn’t started interactively
|
||||
# https://discourse.julialang.org/t/scripting-like-a-julian/50707
|
||||
!isinteractive() && main()
|
||||
#------------------------------------------------------------------------------------------------100
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
88
previousVersion/0.0.05-alpha/src/IronpenGPU.jl
Normal file
88
previousVersion/0.0.05-alpha/src/IronpenGPU.jl
Normal file
@@ -0,0 +1,88 @@
|
||||
module IronpenGPU # this is a parent module
|
||||
|
||||
# export
|
||||
|
||||
|
||||
""" Order by dependencies of each file. The 1st included file must not depend on any other
|
||||
files and each file can only depend on the file included before it.
|
||||
"""
|
||||
|
||||
include("type.jl")
|
||||
using .type # bring type into parent module namespace
|
||||
|
||||
include("snnUtil.jl")
|
||||
using .snnUtil
|
||||
|
||||
include("forward.jl")
|
||||
using .forward
|
||||
|
||||
include("learn.jl")
|
||||
using .learn
|
||||
|
||||
include("interface.jl")
|
||||
using .interface
|
||||
|
||||
|
||||
#------------------------------------------------------------------------------------------------100
|
||||
|
||||
""" version 0.0.5
|
||||
Todo:
|
||||
[DONE] add excitatory/inhabitory matrix
|
||||
[-] add temporal summation in addition to already used spatial summation.
|
||||
CANCELLED, spatial summation every second until membrane potential reach a threshold
|
||||
is in itself a temporal summation.
|
||||
[x] add neuroplasticity
|
||||
[4] implement dormant connection and pruning machanism. the longer the training the longer
|
||||
0 weight stay 0.
|
||||
[] using RL to control learning signal
|
||||
[] consider using Dates.now() instead of timestamp because time_stamp may overflow
|
||||
[] Liquid time constant. training should include adjusting α, neuron membrane potential decay factor
|
||||
which defined by neuron.tau_m formula in type.jl
|
||||
|
||||
Change from version: 0.0.4
|
||||
-
|
||||
|
||||
All features
|
||||
|
||||
"""
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
end # module IronpenGPU
|
||||
894
previousVersion/0.0.05-alpha/src/forward.jl
Normal file
894
previousVersion/0.0.05-alpha/src/forward.jl
Normal file
@@ -0,0 +1,894 @@
|
||||
module forward
|
||||
|
||||
# export
|
||||
|
||||
using Flux, CUDA
|
||||
using GeneralUtils
|
||||
using ..type, ..snnUtil
|
||||
|
||||
#------------------------------------------------------------------------------------------------100
|
||||
|
||||
""" kfn forward
|
||||
input (row, col, batch)
|
||||
"""
|
||||
function (kfn::kfn_1)(input::AbstractArray)
|
||||
|
||||
kfn.timeStep .+= 1
|
||||
|
||||
# what to do at the start of learning round
|
||||
if view(kfn.learningStage, 1)[1] == 1
|
||||
# reset learning params
|
||||
kfn.zit_cumulative .= 0
|
||||
|
||||
kfn.lif_vt .= 0
|
||||
kfn.lif_wRecChange .= 0
|
||||
kfn.lif_epsilonRec .= 0
|
||||
kfn.lif_firingCounter .= 0
|
||||
kfn.lif_refractoryCounter .= 0
|
||||
kfn.lif_zt .= 0
|
||||
|
||||
kfn.alif_vt .= 0
|
||||
kfn.alif_epsilonRec .= 0
|
||||
kfn.alif_epsilonRecA .= 0
|
||||
kfn.alif_wRecChange .= 0
|
||||
kfn.alif_firingCounter .= 0
|
||||
kfn.alif_refractoryCounter .= 0
|
||||
kfn.alif_zt .= 0
|
||||
|
||||
kfn.on_vt .= 0
|
||||
kfn.on_epsilonRec .= 0
|
||||
kfn.on_wOutChange .= 0
|
||||
kfn.on_refractoryCounter .= 0
|
||||
|
||||
kfn.learningStage = [2]
|
||||
end
|
||||
|
||||
# update activation matrix with "lif_zt1" and "alif_zt1" by concatenating
|
||||
# (input, lif_zt1, alif_zt1) to form activation matrix
|
||||
_zit = cat(reshape(input, (size(input, 1), size(input, 2), 1, size(input, 3))),
|
||||
reshape(kfn.lif_zt, (size(input, 1), :, 1, size(input, 3))),
|
||||
reshape(kfn.alif_zt, (size(input, 1), :, 1, size(input, 3))), dims=2)
|
||||
kfn.zit .= reshape(_zit, (size(input, 1), :, size(input, 3)))
|
||||
|
||||
@sync begin
|
||||
@async begin
|
||||
# project 3D kfn zit into 4D lif zit
|
||||
i1, i2, i3, i4 = size(kfn.lif_zit)
|
||||
kfn.lif_zit .= reshape(kfn.zit, (i1, i2, 1, i4)) .* kfn.lif_arrayProjection4d
|
||||
kfn.lif_exInType .= kfn.exInType .* kfn.lif_arrayProjection4d
|
||||
|
||||
lifForward( kfn.lif_zit,
|
||||
kfn.lif_wRec,
|
||||
kfn.lif_vt,
|
||||
kfn.lif_vth,
|
||||
kfn.lif_vRest,
|
||||
kfn.lif_zt4d,
|
||||
kfn.lif_alpha,
|
||||
kfn.lif_phi,
|
||||
kfn.lif_epsilonRec,
|
||||
kfn.lif_refractoryCounter,
|
||||
kfn.lif_refractoryDuration,
|
||||
kfn.lif_gammaPd,
|
||||
kfn.lif_firingCounter,
|
||||
kfn.lif_recSignal,
|
||||
kfn.lif_exInType,
|
||||
kfn.lif_neuronInactivityCounter,
|
||||
kfn.lif_synapticInactivityCounter,
|
||||
)
|
||||
end
|
||||
@async begin
|
||||
# project 3D kfn zit into 4D alif zit
|
||||
i1, i2, i3, i4 = size(kfn.alif_zit)
|
||||
kfn.alif_zit .= reshape(kfn.zit, (i1, i2, 1, i4)) .* kfn.alif_arrayProjection4d
|
||||
kfn.alif_exInType .= kfn.exInType .* kfn.alif_arrayProjection4d
|
||||
|
||||
alifForward(kfn.alif_zit,
|
||||
kfn.alif_wRec,
|
||||
kfn.alif_vt,
|
||||
kfn.alif_vth,
|
||||
kfn.alif_vRest,
|
||||
kfn.alif_zt4d,
|
||||
kfn.alif_alpha,
|
||||
kfn.alif_phi,
|
||||
kfn.alif_epsilonRec,
|
||||
kfn.alif_refractoryCounter,
|
||||
kfn.alif_refractoryDuration,
|
||||
kfn.alif_gammaPd,
|
||||
kfn.alif_firingCounter,
|
||||
kfn.alif_recSignal,
|
||||
kfn.alif_exInType,
|
||||
kfn.alif_neuronInactivityCounter,
|
||||
kfn.alif_synapticInactivityCounter,
|
||||
kfn.alif_epsilonRecA,
|
||||
kfn.alif_a,
|
||||
kfn.alif_avth,
|
||||
kfn.alif_beta,
|
||||
kfn.alif_rho,
|
||||
)
|
||||
end
|
||||
end
|
||||
|
||||
# reduce lif_zt4d and alif_zt4d into lif_zt, alif_zt (4d -> 1d)
|
||||
kfn.lif_zt .= reduce(max, kfn.lif_zt4d, dims=(1,2))
|
||||
kfn.alif_zt .= reduce(max, kfn.alif_zt4d, dims=(1,2))
|
||||
|
||||
# update activation matrix with "lif_zt1" and "alif_zt1" by concatenating
|
||||
# (input, lif_zt1, alif_zt1) to form activation matrix
|
||||
_zit = cat(reshape(input, (size(input, 1), size(input, 2), 1, size(input, 3))),
|
||||
reshape(kfn.lif_zt, (size(input, 1), :, 1, size(input, 3))),
|
||||
reshape(kfn.alif_zt, (size(input, 1), :, 1, size(input, 3))), dims=2)
|
||||
kfn.zit .= reshape(_zit, (size(input, 1), :, size(input, 3)))
|
||||
kfn.zit_cumulative .+= kfn.zit
|
||||
|
||||
# project 3D kfn zit into 4D on zit
|
||||
i1, i2, i3, i4 = size(kfn.on_zit)
|
||||
kfn.on_zit .= reshape(kfn.zit, (i1, i2, 1, i4)) .* kfn.on_arrayProjection4d
|
||||
|
||||
# read out
|
||||
onForward( kfn.on_zit,
|
||||
kfn.on_wOut,
|
||||
kfn.on_vt,
|
||||
kfn.on_vth,
|
||||
kfn.on_vRest,
|
||||
kfn.on_zt4d,
|
||||
kfn.on_alpha,
|
||||
kfn.on_phi,
|
||||
kfn.on_epsilonRec,
|
||||
kfn.on_refractoryCounter,
|
||||
kfn.on_refractoryDuration,
|
||||
kfn.on_gammaPd,
|
||||
kfn.on_firingCounter,
|
||||
kfn.on_recSignal,
|
||||
)
|
||||
# get on_zt4d to on_zt
|
||||
kfn.on_zt .= reduce(max, kfn.on_zt4d, dims=(1,2))
|
||||
logit = reshape(kfn.on_zt, (size(input, 1), :))
|
||||
|
||||
return logit,
|
||||
kfn.zit
|
||||
end
|
||||
|
||||
# gpu launcher
|
||||
function lifForward( zit::CuArray,
|
||||
wRec::CuArray,
|
||||
vt::CuArray,
|
||||
vth::CuArray,
|
||||
vRest::CuArray,
|
||||
zt::CuArray,
|
||||
alpha::CuArray,
|
||||
phi::CuArray,
|
||||
epsilonRec::CuArray,
|
||||
refractoryCounter::CuArray,
|
||||
refractoryDuration::CuArray,
|
||||
gammaPd::CuArray,
|
||||
firingCounter::CuArray,
|
||||
recSignal::CuArray,
|
||||
exInType::CuArray,
|
||||
neuronInactivityCounter::CuArray,
|
||||
synapticInactivityCounter::CuArray,
|
||||
)
|
||||
|
||||
kernel = @cuda launch=false lifForward( zit,
|
||||
wRec,
|
||||
vt,
|
||||
vth,
|
||||
vRest,
|
||||
zt,
|
||||
alpha,
|
||||
phi,
|
||||
epsilonRec,
|
||||
refractoryCounter,
|
||||
refractoryDuration,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
exInType,
|
||||
neuronInactivityCounter,
|
||||
synapticInactivityCounter,
|
||||
GeneralUtils.linear_to_cartesian,
|
||||
)
|
||||
config = launch_configuration(kernel.fun)
|
||||
|
||||
|
||||
# threads to be launched. Since one can't launch exact thread number the kernel needs,
|
||||
# one just launch threads more than this kernel needs then use a guard inside the kernel
|
||||
# to prevent unused threads to access memory.
|
||||
threads = min(1024, config.threads) # depend on gpu. Most NVIDIA gpu has 1024 threads per block
|
||||
|
||||
# total desired threads to launch to gpu. Usually 1 thread per 1 matrix element
|
||||
totalThreads = length(wRec)
|
||||
|
||||
blocks = cld(totalThreads, threads)
|
||||
# println("launching gpu kernel")
|
||||
CUDA.@sync begin
|
||||
kernel( zit,
|
||||
wRec,
|
||||
vt,
|
||||
vth,
|
||||
vRest,
|
||||
zt,
|
||||
alpha,
|
||||
phi,
|
||||
epsilonRec,
|
||||
refractoryCounter,
|
||||
refractoryDuration,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
exInType,
|
||||
neuronInactivityCounter,
|
||||
synapticInactivityCounter,
|
||||
GeneralUtils.linear_to_cartesian; threads, blocks)
|
||||
end
|
||||
end
|
||||
|
||||
# gpu kernel
|
||||
function lifForward( zit,
|
||||
wRec,
|
||||
vt,
|
||||
vth,
|
||||
vRest,
|
||||
zt,
|
||||
alpha,
|
||||
phi,
|
||||
epsilonRec,
|
||||
refractoryCounter,
|
||||
refractoryDuration,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
exInType,
|
||||
neuronInactivityCounter,
|
||||
synapticInactivityCounter,
|
||||
linear_to_cartesian,
|
||||
)
|
||||
i = (blockIdx().x - 1) * blockDim().x + threadIdx().x # gpu threads index
|
||||
|
||||
if i <= length(wRec)
|
||||
# cartesian index
|
||||
i1, i2, i3, i4 = linear_to_cartesian(i, size(wRec))
|
||||
# @cuprintln("gpu thread $i $i1 $i2 $i3 $i4")
|
||||
|
||||
if refractoryCounter[i1,i2,i3,i4] > 0 # refractory period is active
|
||||
refractoryCounter[i1,i2,i3,i4] -= 1
|
||||
recSignal[i1,i2,i3,i4] = 0
|
||||
zt[i1,i2,i3,i4] = 0
|
||||
vt[i1,i2,i3,i4] = alpha[i1,i2,i3,i4] * vt[i1,i2,i3,i4]
|
||||
phi[i1,i2,i3,i4] = 0
|
||||
|
||||
# compute epsilonRec
|
||||
epsilonRec[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * epsilonRec[i1,i2,i3,i4])
|
||||
|
||||
else # refractory period is inactive
|
||||
recSignal[i1,i2,i3,i4] = wRec[i1,i2,i3,i4] * zit[i1,i2,i3,i4] *
|
||||
exInType[i1,i2,i3,i4]
|
||||
vt[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * vt[i1,i2,i3,i4]) +
|
||||
sum(@view(recSignal[:,:,i3,i4]))
|
||||
|
||||
# fires if membrane potential exceed threshold
|
||||
if vt[i1,i2,i3,i4] > vth[i1,i2,i3,i4]
|
||||
zt[i1,i2,i3,i4] = 1
|
||||
refractoryCounter[i1,i2,i3,i4] = refractoryDuration[i1,i2,i3,i4]
|
||||
firingCounter[i1,i2,i3,i4] += 1
|
||||
vt[i1,i2,i3,i4] = vRest[i1,i2,i3,i4]
|
||||
|
||||
# reset counter if neuron fires
|
||||
neuronInactivityCounter[i1,i2,i3,i4] = 10000
|
||||
else
|
||||
zt[i1,i2,i3,i4] = 0
|
||||
neuronInactivityCounter[i1,i2,i3,i4] -= 1
|
||||
end
|
||||
|
||||
# compute phi, there is a difference from lif formula
|
||||
phi[i1,i2,i3,i4] = (gammaPd[i1,i2,i3,i4] / vth[i1,i2,i3,i4]) *
|
||||
max(0, 1 - ((vt[i1,i2,i3,i4] - vth[i1,i2,i3,i4]) / vth[i1,i2,i3,i4]))
|
||||
|
||||
# compute epsilonRec
|
||||
epsilonRec[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * epsilonRec[i1,i2,i3,i4]) +
|
||||
(zit[i1,i2,i3,i4] * !iszero(wRec[i1,i2,i3,i4]))
|
||||
# !iszero indicates synaptic subscription
|
||||
|
||||
# count synaptic inactivity
|
||||
if !iszero(wRec[i1,i2,i3,i4]) # check if this is wRec subscription
|
||||
if !iszero(zit[i1,i2,i3,i4]) # synapse is active, reset counter
|
||||
synapticInactivityCounter[i1,i2,i3,i4] = 10000
|
||||
else # synapse is inactive, counting
|
||||
synapticInactivityCounter[i1,i2,i3,i4] -= 1
|
||||
end
|
||||
end
|
||||
|
||||
end
|
||||
end
|
||||
return nothing
|
||||
end
|
||||
|
||||
# gpu launcher
|
||||
function alifForward( zit::CuArray,
|
||||
wRec::CuArray,
|
||||
vt::CuArray,
|
||||
vth::CuArray,
|
||||
vRest::CuArray,
|
||||
zt::CuArray,
|
||||
alpha::CuArray,
|
||||
phi::CuArray,
|
||||
epsilonRec::CuArray,
|
||||
refractoryCounter::CuArray,
|
||||
refractoryDuration::CuArray,
|
||||
gammaPd::CuArray,
|
||||
firingCounter::CuArray,
|
||||
recSignal::CuArray,
|
||||
exInType::CuArray,
|
||||
neuronInactivityCounter::CuArray,
|
||||
synapticInactivityCounter::CuArray,
|
||||
epsilonRecA::CuArray,
|
||||
a::CuArray,
|
||||
avth::CuArray,
|
||||
beta::CuArray,
|
||||
rho::CuArray,
|
||||
)
|
||||
|
||||
kernel = @cuda launch=false alifForward( zit,
|
||||
wRec,
|
||||
vt,
|
||||
vth,
|
||||
vRest,
|
||||
zt,
|
||||
alpha,
|
||||
phi,
|
||||
epsilonRec,
|
||||
refractoryCounter,
|
||||
refractoryDuration,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
exInType,
|
||||
neuronInactivityCounter,
|
||||
synapticInactivityCounter,
|
||||
epsilonRecA,
|
||||
a,
|
||||
avth,
|
||||
beta,
|
||||
rho,
|
||||
GeneralUtils.linear_to_cartesian,
|
||||
)
|
||||
config = launch_configuration(kernel.fun)
|
||||
|
||||
# threads to be launched. Since one can't launch exact thread number the kernel needs,
|
||||
# one just launch threads more than this kernel needs then use a guard inside the kernel
|
||||
# to prevent unused threads to access memory.
|
||||
threads = min(1024, config.threads) # depend on gpu. Most NVIDIA gpu has 1024 threads per block
|
||||
|
||||
# total desired threads to launch to gpu. Usually 1 thread per 1 matrix element
|
||||
totalThreads = length(wRec)
|
||||
|
||||
blocks = cld(totalThreads, threads)
|
||||
# println("launching gpu kernel")
|
||||
CUDA.@sync begin
|
||||
kernel( zit,
|
||||
wRec,
|
||||
vt,
|
||||
vth,
|
||||
vRest,
|
||||
zt,
|
||||
alpha,
|
||||
phi,
|
||||
epsilonRec,
|
||||
refractoryCounter,
|
||||
refractoryDuration,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
exInType,
|
||||
neuronInactivityCounter,
|
||||
synapticInactivityCounter,
|
||||
epsilonRecA,
|
||||
a,
|
||||
avth,
|
||||
beta,
|
||||
rho,
|
||||
GeneralUtils.linear_to_cartesian; threads, blocks)
|
||||
end
|
||||
end
|
||||
|
||||
# gpu kernel
|
||||
function alifForward( zit,
|
||||
wRec,
|
||||
vt,
|
||||
vth,
|
||||
vRest,
|
||||
zt,
|
||||
alpha,
|
||||
phi,
|
||||
epsilonRec,
|
||||
refractoryCounter,
|
||||
refractoryDuration,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
exInType,
|
||||
neuronInactivityCounter,
|
||||
synapticInactivityCounter,
|
||||
epsilonRecA,
|
||||
a,
|
||||
avth,
|
||||
beta,
|
||||
rho,
|
||||
linear_to_cartesian,
|
||||
)
|
||||
i = (blockIdx().x - 1) * blockDim().x + threadIdx().x # gpu threads index
|
||||
|
||||
if i <= length(wRec)
|
||||
# cartesian index
|
||||
i1, i2, i3, i4 = linear_to_cartesian(i, size(wRec))
|
||||
# @cuprintln("gpu thread $i $i1 $i2 $i3 $i4")
|
||||
|
||||
if refractoryCounter[i1,i2,i3,i4] > 0 # refractory period is active
|
||||
refractoryCounter[i1,i2,i3,i4] -= 1
|
||||
recSignal[i1,i2,i3,i4] = 0
|
||||
zt[i1,i2,i3,i4] = 0
|
||||
vt[i1,i2,i3,i4] = alpha[i1,i2,i3,i4] * vt[i1,i2,i3,i4]
|
||||
phi[i1,i2,i3,i4] = 0
|
||||
a[i1,i2,i3,i4] = rho[i1,i2,i3,i4] * a[i1,i2,i3,i4]
|
||||
|
||||
# compute epsilonRec
|
||||
epsilonRec[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * epsilonRec[i1,i2,i3,i4])
|
||||
|
||||
# compute epsilonRecA use eq.26
|
||||
epsilonRecA[i1,i2,i3,i4] = (rho[i1,i2,i3,i4] *
|
||||
(phi[i1,i2,i3,i4] * epsilonRec[i1,i2,i3,i4]))
|
||||
|
||||
# compute avth
|
||||
avth[i1,i2,i3,i4] = vth[i1,i2,i3,i4] + (beta[i1,i2,i3,i4] * a[i1,i2,i3,i4])
|
||||
|
||||
else # refractory period is inactive
|
||||
recSignal[i1,i2,i3,i4] = wRec[i1,i2,i3,i4] * zit[i1,i2,i3,i4] *
|
||||
exInType[i1,i2,i3,i4]
|
||||
vt[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * vt[i1,i2,i3,i4]) +
|
||||
sum(@view(recSignal[:,:,i3,i4]))
|
||||
|
||||
# compute avth
|
||||
avth[i1,i2,i3,i4] = vth[i1,i2,i3,i4] + (beta[i1,i2,i3,i4] * a[i1,i2,i3,i4])
|
||||
|
||||
# fires if membrane potential exceed threshold
|
||||
if vt[i1,i2,i3,i4] > avth[i1,i2,i3,i4]
|
||||
zt[i1,i2,i3,i4] = 1
|
||||
refractoryCounter[i1,i2,i3,i4] = refractoryDuration[i1,i2,i3,i4]
|
||||
firingCounter[i1,i2,i3,i4] += 1
|
||||
vt[i1,i2,i3,i4] = vRest[i1,i2,i3,i4]
|
||||
a[i1,i2,i3,i4] = (rho[i1,i2,i3,i4] * a[i1,i2,i3,i4]) + 1
|
||||
neuronInactivityCounter[i1,i2,i3,i4] = 10000
|
||||
else
|
||||
zt[i1,i2,i3,i4] = 0
|
||||
a[i1,i2,i3,i4] = (rho[i1,i2,i3,i4] * a[i1,i2,i3,i4])
|
||||
neuronInactivityCounter[i1,i2,i3,i4] -= 1
|
||||
end
|
||||
|
||||
# compute phi, there is a difference from alif formula
|
||||
phi[i1,i2,i3,i4] = (gammaPd[i1,i2,i3,i4] / vth[i1,i2,i3,i4]) *
|
||||
max(0, 1 - ((vt[i1,i2,i3,i4] - vth[i1,i2,i3,i4]) / vth[i1,i2,i3,i4]))
|
||||
|
||||
# compute epsilonRec
|
||||
epsilonRec[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * epsilonRec[i1,i2,i3,i4]) +
|
||||
(zit[i1,i2,i3,i4] * !iszero(wRec[i1,i2,i3,i4]))
|
||||
# compute epsilonRecA use eq.26
|
||||
epsilonRecA[i1,i2,i3,i4] = (rho[i1,i2,i3,i4] *
|
||||
(phi[i1,i2,i3,i4] * epsilonRec[i1,i2,i3,i4])) +
|
||||
(zit[i1,i2,i3,i4] * !iszero(wRec[i1,i2,i3,i4]))
|
||||
|
||||
# count synaptic inactivity
|
||||
if !iszero(wRec[i1,i2,i3,i4]) # check if this is wRec subscription
|
||||
if !iszero(zit[i1,i2,i3,i4]) # synapse is active, reset counter
|
||||
synapticInactivityCounter[i1,i2,i3,i4] = 10000
|
||||
else # synapse is inactive, counting
|
||||
synapticInactivityCounter[i1,i2,i3,i4] -= 1
|
||||
end
|
||||
end
|
||||
end
|
||||
end
|
||||
return nothing
|
||||
end
|
||||
|
||||
# gpu launcher
|
||||
function onForward( zit::CuArray,
|
||||
wOut::CuArray,
|
||||
vt::CuArray,
|
||||
vth::CuArray,
|
||||
vRest::CuArray,
|
||||
zt::CuArray,
|
||||
alpha::CuArray,
|
||||
phi::CuArray,
|
||||
epsilonRec::CuArray,
|
||||
refractoryCounter::CuArray,
|
||||
refractoryDuration::CuArray,
|
||||
gammaPd::CuArray,
|
||||
firingCounter::CuArray,
|
||||
recSignal::CuArray,
|
||||
)
|
||||
|
||||
kernel = @cuda launch=false onForward( zit,
|
||||
wOut,
|
||||
vt,
|
||||
vth,
|
||||
vRest,
|
||||
zt,
|
||||
alpha,
|
||||
phi,
|
||||
epsilonRec,
|
||||
refractoryCounter,
|
||||
refractoryDuration,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
GeneralUtils.linear_to_cartesian,
|
||||
)
|
||||
config = launch_configuration(kernel.fun)
|
||||
|
||||
# threads to be launched. Since one can't launch exact thread number the kernel needs,
|
||||
# one just launch threads more than this kernel needs then use a guard inside the kernel
|
||||
# to prevent unused threads to access memory.
|
||||
threads = min(1024, config.threads) # depend on gpu. Most NVIDIA gpu has 1024 threads per block
|
||||
|
||||
# total desired threads to launch to gpu. Usually 1 thread per 1 matrix element
|
||||
totalThreads = length(wOut)
|
||||
|
||||
blocks = cld(totalThreads, threads)
|
||||
# println("launching gpu kernel")
|
||||
CUDA.@sync begin
|
||||
kernel( zit,
|
||||
wOut,
|
||||
vt,
|
||||
vth,
|
||||
vRest,
|
||||
zt,
|
||||
alpha,
|
||||
phi,
|
||||
epsilonRec,
|
||||
refractoryCounter,
|
||||
refractoryDuration,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
GeneralUtils.linear_to_cartesian; threads, blocks)
|
||||
end
|
||||
end
|
||||
|
||||
# gpu kernel
|
||||
function onForward( zit,
|
||||
wOut,
|
||||
vt,
|
||||
vth,
|
||||
vRest,
|
||||
zt,
|
||||
alpha,
|
||||
phi,
|
||||
epsilonRec,
|
||||
refractoryCounter,
|
||||
refractoryDuration,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
linear_to_cartesian,
|
||||
)
|
||||
i = (blockIdx().x - 1) * blockDim().x + threadIdx().x # gpu threads index
|
||||
|
||||
if i <= length(wOut)
|
||||
# cartesian index
|
||||
i1, i2, i3, i4 = linear_to_cartesian(i, size(wOut))
|
||||
# @cuprintln("gpu thread $i $i1 $i2 $i3 $i4")
|
||||
|
||||
if refractoryCounter[i1,i2,i3,i4] > 0 # refractory period is active
|
||||
refractoryCounter[i1,i2,i3,i4] -= 1
|
||||
recSignal[i1,i2,i3,i4] = 0
|
||||
zt[i1,i2,i3,i4] = 0
|
||||
vt[i1,i2,i3,i4] = alpha[i1,i2,i3,i4] * vt[i1,i2,i3,i4]
|
||||
phi[i1,i2,i3,i4] = 0
|
||||
|
||||
# compute epsilonRec
|
||||
epsilonRec[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * epsilonRec[i1,i2,i3,i4])
|
||||
|
||||
else # refractory period is inactive
|
||||
recSignal[i1,i2,i3,i4] = zit[i1,i2,i3,i4] * wOut[i1,i2,i3,i4]
|
||||
vt[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * vt[i1,i2,i3,i4]) + sum(@view(recSignal[:,:,i3,i4]))
|
||||
|
||||
# fires if membrane potential exceed threshold
|
||||
if vt[i1,i2,i3,i4] > vth[i1,i2,i3,i4]
|
||||
zt[i1,i2,i3,i4] = 1
|
||||
refractoryCounter[i1,i2,i3,i4] = refractoryDuration[i1,i2,i3,i4]
|
||||
firingCounter[i1,i2,i3,i4] += 1
|
||||
vt[i1,i2,i3,i4] = vRest[i1,i2,i3,i4]
|
||||
else
|
||||
zt[i1,i2,i3,i4] = 0
|
||||
end
|
||||
|
||||
# compute phi, there is a difference from on formula
|
||||
phi[i1,i2,i3,i4] = (gammaPd[i1,i2,i3,i4] / vth[i1,i2,i3,i4]) *
|
||||
max(0, 1 - ((vt[i1,i2,i3,i4] - vth[i1,i2,i3,i4]) / vth[i1,i2,i3,i4]))
|
||||
|
||||
# compute epsilonRec
|
||||
epsilonRec[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * epsilonRec[i1,i2,i3,i4]) +
|
||||
(zit[i1,i2,i3,i4] * !iszero(wOut[i1,i2,i3,i4]))
|
||||
end
|
||||
end
|
||||
return nothing
|
||||
end
|
||||
|
||||
function lifForward(kfn_zit::Array{T},
|
||||
zit::Array{T},
|
||||
wRec::Array{T},
|
||||
vt0::Array{T},
|
||||
vt1::Array{T},
|
||||
vth::Array{T},
|
||||
vRest::Array{T},
|
||||
zt1::Array{T},
|
||||
alpha::Array{T},
|
||||
phi::Array{T},
|
||||
epsilonRec::Array{T},
|
||||
refractoryCounter::Array{T},
|
||||
refractoryDuration::Array{T},
|
||||
gammaPd::Array{T},
|
||||
firingCounter::Array{T},
|
||||
arrayProjection4d::Array{T},
|
||||
recSignal::Array{T},
|
||||
decayed_vt0::Array{T},
|
||||
decayed_epsilonRec::Array{T},
|
||||
vt1_diff_vth::Array{T},
|
||||
vt1_diff_vth_div_vth::Array{T},
|
||||
gammaPd_div_vth::Array{T},
|
||||
phiActivation::Array{T},
|
||||
) where T<:Number
|
||||
|
||||
# project 3D kfn zit into 4D lif zit
|
||||
i1, i2, i3, i4 = size(alif_wRec)
|
||||
lif_zit .= reshape(kfn_zit, (i1, i2, 1, i4)) .* lif_arrayProjection4d
|
||||
|
||||
for j in 1:size(wRec, 4), i in 1:size(wRec, 3) # compute along neurons axis of every batch
|
||||
if sum(@view(refractoryCounter[:,:,i,j])) > 0 # refractory period is active
|
||||
@. @views refractoryCounter[:,:,i,j] -= 1
|
||||
@. @views zt1[:,:,i,j] = 0
|
||||
@. @views vt1[:,:,i,j] = alpha[:,:,i,j] * vt0[:,:,i,j]
|
||||
@. @views phi[:,:,i,j] = 0
|
||||
|
||||
# compute epsilonRec
|
||||
@. @views decayed_epsilonRec[:,:,i,j] = alpha[:,:,i,j] * epsilonRec[:,:,i,j]
|
||||
@. @views epsilonRec[:,:,i,j] = decayed_epsilonRec[:,:,i,j]
|
||||
else # refractory period is inactive
|
||||
@. @views recSignal[:,:,i,j] = zit[:,:,i,j] * wRec[:,:,i,j]
|
||||
@. @views decayed_vt0[:,:,i,j] = alpha[:,:,i,j] * vt0[:,:,i,j]
|
||||
@view(vt1[:,:,i,j]) .= @view(decayed_vt0[:,:,i,j]) .+ sum(@view(recSignal[:,:,i,j]))
|
||||
|
||||
if sum(@view(vt1[:,:,i,j])) > sum(@view(vth[:,:,i,j]))
|
||||
@. @views zt1[:,:,i,j] = 1
|
||||
@. @views refractoryCounter[:,:,i,j] = refractoryDuration[:,:,i,j]
|
||||
@. @views firingCounter[:,:,i,j] += 1
|
||||
@. @views vt1[:,:,i,j] = vRest[:,:,i,j]
|
||||
else
|
||||
@. @views zt1[:,:,i,j] = 0
|
||||
end
|
||||
|
||||
# compute phi, there is a difference from alif formula
|
||||
@. @views gammaPd_div_vth[:,:,i,j] = gammaPd[:,:,i,j] / vth[:,:,i,j]
|
||||
@. @views vt1_diff_vth[:,:,i,j] = vt1[:,:,i,j] - vth[:,:,i,j]
|
||||
@. @views vt1_diff_vth_div_vth[:,:,i,j] = vt1_diff_vth[:,:,i,j] / vth[:,:,i,j]
|
||||
@view(phiActivation[:,:,i,j]) .= max(0, 1 - sum(@view(vt1_diff_vth_div_vth[:,:,i,j])))
|
||||
@. @views phi[:,:,i,j] = gammaPd_div_vth[:,:,i,j] * phiActivation[:,:,i,j]
|
||||
|
||||
# compute epsilonRec
|
||||
@. @views decayed_epsilonRec[:,:,i,j] = alpha[:,:,i,j] * epsilonRec[:,:,i,j]
|
||||
@. @views epsilonRec[:,:,i,j] = decayed_epsilonRec[:,:,i,j] + zit[:,:,i,j]
|
||||
end
|
||||
end
|
||||
end
|
||||
|
||||
function alifForward(zit::Array{T},
|
||||
wRec::Array{T},
|
||||
vt0::Array{T},
|
||||
vt1::Array{T},
|
||||
vth::Array{T},
|
||||
vRest::Array{T},
|
||||
zt1::Array{T},
|
||||
alpha::Array{T},
|
||||
phi::Array{T},
|
||||
epsilonRec::Array{T},
|
||||
refractoryCounter::Array{T},
|
||||
refractoryDuration::Array{T},
|
||||
gammaPd::Array{T},
|
||||
firingCounter::Array{T},
|
||||
recSignal::Array{T},
|
||||
decayed_vt0::Array{T},
|
||||
decayed_epsilonRec::Array{T},
|
||||
vt1_diff_vth::Array{T},
|
||||
vt1_diff_vth_div_vth::Array{T},
|
||||
gammaPd_div_vth::Array{T},
|
||||
phiActivation::Array{T},
|
||||
|
||||
epsilonRecA::Array{T},
|
||||
avth::Array{T},
|
||||
a::Array{T},
|
||||
beta::Array{T},
|
||||
rho::Array{T},
|
||||
phi_x_epsilonRec::Array{T},
|
||||
phi_x_beta::Array{T},
|
||||
rho_diff_phi_x_beta::Array{T},
|
||||
rho_div_phi_x_beta_x_epsilonRecA::Array{T},
|
||||
beta_x_a::Array{T},
|
||||
) where T<:Number
|
||||
|
||||
for j in 1:size(wRec, 4), i in 1:size(wRec, 3) # compute along neurons axis of every batch
|
||||
if sum(@view(refractoryCounter[:,:,i,j])) > 0 # refractory period is active
|
||||
@. @views refractoryCounter[:,:,i,j] -= 1
|
||||
@. @views zt1[:,:,i,j] = 0
|
||||
@. @views vt1[:,:,i,j] = alpha[:,:,i,j] * vt0[:,:,i,j]
|
||||
@. @views phi[:,:,i,j] = 0
|
||||
@. @views a[:,:,i,j] = rho[:,:,i,j] * a[:,:,i,j]
|
||||
|
||||
# compute epsilonRec
|
||||
@. @views decayed_epsilonRec[:,:,i,j] = alpha[:,:,i,j] * epsilonRec[:,:,i,j]
|
||||
@. @views epsilonRec[:,:,i,j] = decayed_epsilonRec[:,:,i,j]
|
||||
|
||||
# compute epsilonRecA
|
||||
@. @views phi_x_epsilonRec[:,:,i,j] = phi[:,:,i,j] * epsilonRec[:,:,i,j]
|
||||
@. @views phi_x_beta[:,:,i,j] = phi[:,:,i,j] * beta[:,:,i,j]
|
||||
@. @views rho_diff_phi_x_beta[:,:,i,j] = rho[:,:,i,j] - phi_x_beta[:,:,i,j]
|
||||
@. @views rho_div_phi_x_beta_x_epsilonRecA[:,:,i,j] = rho_diff_phi_x_beta[:,:,i,j] * epsilonRecA[:,:,i,j]
|
||||
@. @views epsilonRecA[:,:,i,j] = phi_x_epsilonRec[:,:,i,j] + rho_div_phi_x_beta_x_epsilonRecA[:,:,i,j]
|
||||
|
||||
# compute avth
|
||||
@. @views beta_x_a[:,:,i,j] = beta[:,:,i,j] * a[:,:,i,j]
|
||||
@. @views avth[:,:,i,j] = vth[:,:,i,j] + beta_x_a[:,:,i,j]
|
||||
|
||||
else # refractory period is inactive
|
||||
@. @views recSignal[:,:,i,j] = zit[:,:,i,j] * wRec[:,:,i,j]
|
||||
@. @views decayed_vt0[:,:,i,j] = alpha[:,:,i,j] * vt0[:,:,i,j]
|
||||
@view(vt1[:,:,i,j]) .= @view(decayed_vt0[:,:,i,j]) .+ sum(@view(recSignal[:,:,i,j]))
|
||||
|
||||
# compute avth
|
||||
@. @views beta_x_a[:,:,i,j] = beta[:,:,i,j] * a[:,:,i,j]
|
||||
@. @views avth[:,:,i,j] = vth[:,:,i,j] + beta_x_a[:,:,i,j]
|
||||
|
||||
if sum(@view(vt1[:,:,i,j])) > sum(@view(avth[:,:,i,j]))
|
||||
@. @views zt1[:,:,i,j] = 1
|
||||
@. @views refractoryCounter[:,:,i,j] = refractoryDuration[:,:,i,j]
|
||||
@. @views firingCounter[:,:,i,j] += 1
|
||||
@. @views vt1[:,:,i,j] = vRest[:,:,i,j]
|
||||
@. @views a[:,:,i,j] = rho[:,:,i,j] * a[:,:,i,j]
|
||||
@. @views a[:,:,i,j] = a[:,:,i,j] += 1
|
||||
else
|
||||
@. @views zt1[:,:,i,j] = 0
|
||||
@. @views a[:,:,i,j] = rho[:,:,i,j] * a[:,:,i,j]
|
||||
end
|
||||
|
||||
# compute phi, there is a difference from alif formula
|
||||
@. @views gammaPd_div_vth[:,:,i,j] = gammaPd[:,:,i,j] / vth[:,:,i,j]
|
||||
@. @views vt1_diff_vth[:,:,i,j] = vt1[:,:,i,j] - vth[:,:,i,j]
|
||||
@. @views vt1_diff_vth_div_vth[:,:,i,j] = vt1_diff_vth[:,:,i,j] / vth[:,:,i,j]
|
||||
@view(phiActivation[:,:,i,j]) .= max(0, 1 - sum(@view(vt1_diff_vth_div_vth[:,:,i,j])))
|
||||
@. @views phi[:,:,i,j] = gammaPd_div_vth[:,:,i,j] * phiActivation[:,:,i,j]
|
||||
|
||||
# compute epsilonRec
|
||||
@. @views decayed_epsilonRec[:,:,i,j] = alpha[:,:,i,j] * epsilonRec[:,:,i,j]
|
||||
@. @views epsilonRec[:,:,i,j] = decayed_epsilonRec[:,:,i,j] + zit[:,:,i,j]
|
||||
|
||||
# compute epsilonRecA
|
||||
@. @views phi_x_epsilonRec[:,:,i,j] = phi[:,:,i,j] * epsilonRec[:,:,i,j]
|
||||
@. @views phi_x_beta[:,:,i,j] = phi[:,:,i,j] * beta[:,:,i,j]
|
||||
@. @views rho_diff_phi_x_beta[:,:,i,j] = rho[:,:,i,j] - phi_x_beta[:,:,i,j]
|
||||
@. @views rho_div_phi_x_beta_x_epsilonRecA[:,:,i,j] = rho_diff_phi_x_beta[:,:,i,j] * epsilonRecA[:,:,i,j]
|
||||
@. @views epsilonRecA[:,:,i,j] = phi_x_epsilonRec[:,:,i,j] + rho_div_phi_x_beta_x_epsilonRecA[:,:,i,j]
|
||||
end
|
||||
end
|
||||
end
|
||||
|
||||
function onForward(kfn_zit::Array{T},
|
||||
zit::Array{T},
|
||||
wOut::Array{T},
|
||||
vt0::Array{T},
|
||||
vt1::Array{T},
|
||||
vth::Array{T},
|
||||
vRest::Array{T},
|
||||
zt1::Array{T},
|
||||
alpha::Array{T},
|
||||
phi::Array{T},
|
||||
epsilonRec::Array{T},
|
||||
refractoryCounter::Array{T},
|
||||
refractoryDuration::Array{T},
|
||||
gammaPd::Array{T},
|
||||
firingCounter::Array{T},
|
||||
arrayProjection4d::Array{T},
|
||||
recSignal::Array{T},
|
||||
decayed_vt0::Array{T},
|
||||
decayed_epsilonRec::Array{T},
|
||||
vt1_diff_vth::Array{T},
|
||||
vt1_diff_vth_div_vth::Array{T},
|
||||
gammaPd_div_vth::Array{T},
|
||||
phiActivation::Array{T},
|
||||
) where T<:Number
|
||||
|
||||
# project 3D kfn zit into 4D lif zit
|
||||
zit .= reshape(kfn_zit,
|
||||
(size(wOut, 1), size(wOut, 2), 1, size(wOut, 4))) .* arrayProjection4d
|
||||
|
||||
for j in 1:size(wOut, 4), i in 1:size(wOut, 3) # compute along neurons axis of every batch
|
||||
if sum(@view(refractoryCounter[:,:,i,j])) > 0 # refractory period is active
|
||||
@. @views refractoryCounter[:,:,i,j] -= 1
|
||||
@. @views zt1[:,:,i,j] = 0
|
||||
@. @views vt1[:,:,i,j] = alpha[:,:,i,j] * vt0[:,:,i,j]
|
||||
@. @views phi[:,:,i,j] = 0
|
||||
|
||||
# compute epsilonRec
|
||||
@. @views decayed_epsilonRec[:,:,i,j] = alpha[:,:,i,j] * epsilonRec[:,:,i,j]
|
||||
@. @views epsilonRec[:,:,i,j] = decayed_epsilonRec[:,:,i,j]
|
||||
else # refractory period is inactive
|
||||
@. @views recSignal[:,:,i,j] = zit[:,:,i,j] * wOut[:,:,i,j]
|
||||
@. @views decayed_vt0[:,:,i,j] = alpha[:,:,i,j] * vt0[:,:,i,j]
|
||||
@view(vt1[:,:,i,j]) .= @view(decayed_vt0[:,:,i,j]) .+ sum(@view(recSignal[:,:,i,j]))
|
||||
|
||||
if sum(@view(vt1[:,:,i,j])) > sum(@view(vth[:,:,i,j]))
|
||||
@. @views zt1[:,:,i,j] = 1
|
||||
@. @views refractoryCounter[:,:,i,j] = refractoryDuration[:,:,i,j]
|
||||
@. @views firingCounter[:,:,i,j] += 1
|
||||
@. @views vt1[:,:,i,j] = vRest[:,:,i,j]
|
||||
else
|
||||
@. @views zt1[:,:,i,j] = 0
|
||||
end
|
||||
|
||||
# compute phi, there is a difference from alif formula
|
||||
@. @views gammaPd_div_vth[:,:,i,j] = gammaPd[:,:,i,j] / vth[:,:,i,j]
|
||||
@. @views vt1_diff_vth[:,:,i,j] = vt1[:,:,i,j] - vth[:,:,i,j]
|
||||
@. @views vt1_diff_vth_div_vth[:,:,i,j] = vt1_diff_vth[:,:,i,j] / vth[:,:,i,j]
|
||||
@view(phiActivation[:,:,i,j]) .= max(0, 1 - sum(@view(vt1_diff_vth_div_vth[:,:,i,j])))
|
||||
@. @views phi[:,:,i,j] = gammaPd_div_vth[:,:,i,j] * phiActivation[:,:,i,j]
|
||||
|
||||
# compute epsilonRec
|
||||
@. @views decayed_epsilonRec[:,:,i,j] = alpha[:,:,i,j] * epsilonRec[:,:,i,j]
|
||||
@. @views epsilonRec[:,:,i,j] = decayed_epsilonRec[:,:,i,j] + zit[:,:,i,j]
|
||||
end
|
||||
end
|
||||
end
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
end # module
|
||||
87
previousVersion/0.0.05-alpha/src/interface.jl
Normal file
87
previousVersion/0.0.05-alpha/src/interface.jl
Normal file
@@ -0,0 +1,87 @@
|
||||
module interface
|
||||
|
||||
|
||||
# export
|
||||
|
||||
# using Flux, CUDA
|
||||
|
||||
#------------------------------------------------------------------------------------------------100
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
end # module
|
||||
466
previousVersion/0.0.05-alpha/src/learn.jl
Normal file
466
previousVersion/0.0.05-alpha/src/learn.jl
Normal file
@@ -0,0 +1,466 @@
|
||||
module learn
|
||||
|
||||
export learn!, compute_paramsChange!
|
||||
|
||||
using Statistics, Random, LinearAlgebra, JSON3, Flux, CUDA, Dates
|
||||
using GeneralUtils
|
||||
using ..type, ..snnUtil
|
||||
|
||||
#------------------------------------------------------------------------------------------------100
|
||||
|
||||
function compute_paramsChange!(kfn::kfn_1, modelError, outputError)
|
||||
# modelError = reshape(modelError, (1,1,1,:)) # (1,1,1,batch)
|
||||
modelError = reshape(modelError, (1,1,:, size(modelError, 2)))
|
||||
modelError = sum(modelError, dims=3)
|
||||
|
||||
lifComputeParamsChange!(kfn.timeStep,
|
||||
kfn.lif_phi,
|
||||
kfn.lif_epsilonRec,
|
||||
kfn.lif_eta,
|
||||
kfn.lif_eRec,
|
||||
kfn.lif_wRec,
|
||||
kfn.lif_wRecChange,
|
||||
kfn.on_wOut,
|
||||
kfn.lif_firingCounter,
|
||||
kfn.lif_firingTargetFrequency,
|
||||
kfn.lif_arrayProjection4d,
|
||||
kfn.lif_error,
|
||||
modelError,
|
||||
|
||||
kfn.inputSize,
|
||||
)
|
||||
|
||||
alifComputeParamsChange!(kfn.timeStep,
|
||||
kfn.alif_phi,
|
||||
kfn.alif_epsilonRec,
|
||||
kfn.alif_eta,
|
||||
kfn.alif_eRec,
|
||||
kfn.alif_wRec,
|
||||
kfn.alif_wRecChange,
|
||||
kfn.on_wOut,
|
||||
kfn.alif_firingCounter,
|
||||
kfn.alif_firingTargetFrequency,
|
||||
kfn.alif_arrayProjection4d,
|
||||
kfn.alif_error,
|
||||
modelError,
|
||||
|
||||
kfn.alif_epsilonRecA,
|
||||
kfn.alif_beta,
|
||||
)
|
||||
|
||||
onComputeParamsChange!(kfn.on_phi,
|
||||
kfn.on_epsilonRec,
|
||||
kfn.on_eta,
|
||||
kfn.on_eRec,
|
||||
kfn.on_wOutChange,
|
||||
kfn.on_arrayProjection4d,
|
||||
kfn.on_error,
|
||||
outputError,
|
||||
)
|
||||
# error("DEBUG -> kfn compute_paramsChange! $(Dates.now())")
|
||||
end
|
||||
|
||||
function lifComputeParamsChange!( timeStep::CuArray,
|
||||
phi::CuArray,
|
||||
epsilonRec::CuArray,
|
||||
eta::CuArray,
|
||||
eRec::CuArray,
|
||||
wRec::CuArray,
|
||||
wRecChange::CuArray,
|
||||
wOut::CuArray,
|
||||
firingCounter::CuArray,
|
||||
firingTargetFrequency::CuArray,
|
||||
arrayProjection4d::CuArray,
|
||||
nError::CuArray,
|
||||
modelError::CuArray,
|
||||
|
||||
inputSize::CuArray,
|
||||
)
|
||||
# Bₖⱼ in paper, sum() to get each neuron's total wOut weight,
|
||||
# use absolute because only magnitude is needed
|
||||
wOutSum_all = reshape( abs.(sum(wOut, dims=3)), (1,1,:, size(wOut, 4)) ) # (1,1,allNeuron,batch)
|
||||
|
||||
# get only each lif neuron's wOut, leaving out other neuron's wOut
|
||||
startIndex = prod(inputSize) +1
|
||||
stopIndex = startIndex + size(wRec, 3) -1
|
||||
wOutSum = @view(wOutSum_all[1,1, startIndex:stopIndex, :])
|
||||
wOutSum = reshape(wOutSum, (1, 1, size(wOutSum, 1), size(wOutSum, 2))) # (1,1,n,batch)
|
||||
|
||||
# nError a.k.a. learning signal use dopamine concept,
|
||||
# this neuron receive summed error signal (modelError)
|
||||
nError .= (modelError .* wOutSum) .* arrayProjection4d
|
||||
eRec .= phi .* epsilonRec
|
||||
wRecChange .+= (-eta .* nError .* eRec)
|
||||
|
||||
# frequency regulator
|
||||
wRecChange .+= 0.001 .* ((firingTargetFrequency - (firingCounter./timeStep)) ./ timeStep) .*
|
||||
eta .* eRec
|
||||
|
||||
# if sum(timeStep) == 785
|
||||
# epsilonRec_cpu = epsilonRec |> cpu
|
||||
# println("modelError ", modelError)
|
||||
# println("")
|
||||
# wchange = (-eta .* nError .* eRec) |> cpu
|
||||
# println("wchange 5 1 ", wchange[:,:,5,1])
|
||||
# println("")
|
||||
# println("wchange 5 2 ", wchange[:,:,5,2])
|
||||
# println("")
|
||||
# println("epsilonRec 5 1 ", epsilonRec_cpu[:,:,5,1])
|
||||
# println("")
|
||||
# println("epsilonRec 5 2 ", epsilonRec_cpu[:,:,5,2])
|
||||
# println("")
|
||||
# error("DEBUG lifComputeParamsChange!")
|
||||
# end
|
||||
|
||||
# reset epsilonRec
|
||||
epsilonRec .= 0
|
||||
end
|
||||
|
||||
function alifComputeParamsChange!( timeStep::CuArray,
|
||||
phi::CuArray,
|
||||
epsilonRec::CuArray,
|
||||
eta::CuArray,
|
||||
eRec::CuArray,
|
||||
wRec::CuArray,
|
||||
wRecChange::CuArray,
|
||||
wOut::CuArray,
|
||||
firingCounter::CuArray,
|
||||
firingTargetFrequency::CuArray,
|
||||
arrayProjection4d::CuArray,
|
||||
nError::CuArray,
|
||||
modelError::CuArray,
|
||||
|
||||
epsilonRecA::CuArray,
|
||||
beta::CuArray
|
||||
)
|
||||
|
||||
# Bₖⱼ in paper, sum() to get each neuron's total wOut weight,
|
||||
# use absolute because only magnitude is needed
|
||||
wOutSum_all = reshape( abs.(sum(wOut, dims=3)), (1,1,:, size(wOut, 4)) ) # (1,1,allNeuron,batch)
|
||||
|
||||
# get only each lif neuron's wOut, leaving out other neuron's wOut
|
||||
wOutSum = @view(wOutSum_all[1,1, end-size(wRec, 3)+1:end, :])
|
||||
wOutSum = reshape(wOutSum, (1, 1, size(wOutSum, 1), size(wOutSum, 2))) # (1,1,n,batch)
|
||||
|
||||
# nError a.k.a. learning signal use dopamine concept,
|
||||
# this neuron receive summed error signal (modelError)
|
||||
nError .= (modelError .* wOutSum) .* arrayProjection4d
|
||||
eRec .= phi .* (epsilonRec .- (beta .* epsilonRecA)) # use eq. 25
|
||||
wRecChange .+= (-eta .* nError .* eRec)
|
||||
|
||||
# frequency regulator
|
||||
wRecChange .+= 0.001 .* ((firingTargetFrequency - (firingCounter./timeStep)) ./ timeStep) .*
|
||||
eta .* eRec
|
||||
|
||||
# reset epsilonRec
|
||||
epsilonRec .= 0
|
||||
epsilonRecA .= 0
|
||||
|
||||
# error("DEBUG -> alifComputeParamsChange! $(Dates.now())")
|
||||
end
|
||||
|
||||
function onComputeParamsChange!(phi::CuArray,
|
||||
epsilonRec::CuArray,
|
||||
eta::CuArray,
|
||||
eRec::CuArray,
|
||||
wOutChange::CuArray,
|
||||
arrayProjection4d::CuArray,
|
||||
nError::CuArray,
|
||||
outputError::CuArray # outputError is output neuron's error
|
||||
)
|
||||
|
||||
eRec .= phi .* epsilonRec
|
||||
nError .= reshape(outputError, (1, 1, :, size(outputError, 2))) .* arrayProjection4d
|
||||
wOutChange .+= (-eta .* nError .* eRec)
|
||||
|
||||
# reset epsilonRec
|
||||
epsilonRec .= 0
|
||||
|
||||
# error("DEBUG -> onComputeParamsChange! $(Dates.now())")
|
||||
end
|
||||
|
||||
function lifComputeParamsChange!( phi::AbstractArray,
|
||||
epsilonRec::AbstractArray,
|
||||
eta::AbstractArray,
|
||||
wRec::AbstractArray,
|
||||
wRecChange::AbstractArray,
|
||||
wOut::AbstractArray,
|
||||
modelError::AbstractArray)
|
||||
d1, d2, d3, d4 = size(epsilonRec)
|
||||
# Bₖⱼ in paper, sum() to get each neuron's total wOut weight
|
||||
wOutSum = reshape(sum(wOut, dims=3), (d1, :, d4))
|
||||
|
||||
for j in 1:d4, i in 1:d3 # compute along neurons axis of every batch
|
||||
# how much error of this neuron 1-spike causing each output neuron's error
|
||||
|
||||
view(wRecChange, :, :, i, j) .+= (-1 * view(eta, :, :, i, j)[1]) .*
|
||||
# eRec
|
||||
(
|
||||
(view(phi, :, :, i, j)[1] .* view(epsilonRec, :, :, i, j)) .*
|
||||
# nError a.k.a. learning signal
|
||||
(
|
||||
view(modelError, :, j)[1] * # dopamine concept, this neuron receive summed error signal
|
||||
# RSNN neuron's total wOut weight (neuron synaptic subscription .* wOutSum)
|
||||
view(wOutSum, :, :, j)[i]
|
||||
)
|
||||
)
|
||||
end
|
||||
end
|
||||
|
||||
function alifComputeParamsChange!( phi::AbstractArray,
|
||||
epsilonRec::AbstractArray,
|
||||
epsilonRecA::AbstractArray,
|
||||
eta::AbstractArray,
|
||||
wRec::AbstractArray,
|
||||
wRecChange::AbstractArray,
|
||||
beta::AbstractArray,
|
||||
wOut::AbstractArray,
|
||||
modelError::AbstractArray)
|
||||
d1, d2, d3, d4 = size(epsilonRec)
|
||||
|
||||
# Bₖⱼ in paper, sum() to get each neuron's total wOut weight
|
||||
wOutSum = reshape(sum(wOut, dims=3), (d1, :, d4))
|
||||
|
||||
for j in 1:d4, i in 1:d3 # compute along neurons axis of every batch
|
||||
# how much error of this neuron 1-spike causing each output neuron's error
|
||||
|
||||
view(wRecChange, :, :, i, j) .+= (-1 * view(eta, :, :, i, j)[1]) .*
|
||||
# eRec
|
||||
(
|
||||
# eRec_v
|
||||
(view(phi, :, :, i, j)[1] .* view(epsilonRec, :, :, i, j)) .+
|
||||
# eRec_a
|
||||
((view(phi, :, :, i, j)[1] * view(beta, :, :, i, j)[1]) .*
|
||||
view(epsilonRecA, :, :, i, j))
|
||||
) .*
|
||||
# nError a.k.a. learning signal
|
||||
(
|
||||
view(modelError, :, j)[1] *
|
||||
# RSNN neuron's total wOut weight (neuron synaptic subscription .* wOutSum)
|
||||
view(wOutSum, :, :, j)[i]
|
||||
# sum(GeneralUtils.isNotEqual.(view(wRec, :, :, i, j), 0) .*
|
||||
# view(wOutSum, :, :, j))
|
||||
)
|
||||
end
|
||||
end
|
||||
|
||||
function onComputeParamsChange!(phi::AbstractArray,
|
||||
epsilonRec::AbstractArray,
|
||||
eta::AbstractArray,
|
||||
wOutChange::AbstractArray,
|
||||
outputError::AbstractArray)
|
||||
d1, d2, d3, d4 = size(epsilonRec)
|
||||
|
||||
for j in 1:d4, i in 1:d3 # compute along neurons axis of every batch
|
||||
# how much error of this neuron 1-spike causing each output neuron's error
|
||||
|
||||
view(wOutChange, :, :, i, j) .+= (-1 * view(eta, :, :, i, j)[1]) .*
|
||||
# eRec
|
||||
(
|
||||
(view(phi, :, :, i, j)[1] .* view(epsilonRec, :, :, i, j)) .*
|
||||
# nError a.k.a. learning signal, output neuron receives error of its own answer - correct answer.
|
||||
view(outputError, :, j)[i]
|
||||
)
|
||||
end
|
||||
end
|
||||
|
||||
function learn!(kfn::kfn_1, device=cpu)
|
||||
# lif learn
|
||||
lifLearn!(kfn.lif_wRec,
|
||||
kfn.lif_wRecChange,
|
||||
kfn.lif_arrayProjection4d,
|
||||
kfn.lif_neuronInactivityCounter,
|
||||
kfn.lif_synapticConnectionNumber,
|
||||
kfn.zit_cumulative,
|
||||
device)
|
||||
|
||||
# alif learn
|
||||
alifLearn!(kfn.alif_wRec,
|
||||
kfn.alif_wRecChange,
|
||||
kfn.alif_arrayProjection4d,
|
||||
kfn.alif_neuronInactivityCounter,
|
||||
kfn.alif_synapticConnectionNumber,
|
||||
kfn.zit_cumulative,
|
||||
device)
|
||||
|
||||
# on learn
|
||||
onLearn!(kfn.on_wOut,
|
||||
kfn.on_wOutChange,
|
||||
kfn.on_arrayProjection4d)
|
||||
|
||||
# wrap up learning session
|
||||
if kfn.learningStage == [3]
|
||||
kfn.learningStage = [0]
|
||||
end
|
||||
# error("DEBUG -> kfn learn! $(Dates.now())")
|
||||
end
|
||||
|
||||
function lifLearn!(wRec,
|
||||
wRecChange,
|
||||
arrayProjection4d,
|
||||
inactivityCounter,
|
||||
synapticConnectionNumber,
|
||||
zit_cumulative,
|
||||
device)
|
||||
# merge learning weight with average learning weight of all batch
|
||||
wRec .+= (sum(wRecChange, dims=4) ./ (size(wRec, 4))) .* arrayProjection4d
|
||||
|
||||
wRec_cpu = wRec |> cpu
|
||||
wRec_cpu = wRec_cpu[:,:,:,1] # since every batch has the same neuron wRec, (row, col, n)
|
||||
inactivityCounter_cpu = inactivityCounter |> cpu
|
||||
inactivityCounter_cpu = inactivityCounter_cpu[:,:,:,1] # (row, col, n)
|
||||
zit_cumulative_cpu = zit_cumulative |> cpu
|
||||
zit_cumulative_cpu = zit_cumulative_cpu[:,:,1] # (row, col)
|
||||
|
||||
# weak / negative synaptic connection will get randomed in neuroplasticity()
|
||||
wRec_cpu = GeneralUtils.replaceBetween.(wRec_cpu, 0.0, 0.1, -1.0) # mark with -1.0
|
||||
|
||||
# synaptic connection that has no inactivity will get randomed in neuroplasticity()
|
||||
GeneralUtils.replace_elements!(inactivityCounter_cpu, 0.0, wRec_cpu, -1.0)
|
||||
# reset lif_inactivity elements to 10000
|
||||
GeneralUtils.replace_elements!(inactivityCounter_cpu, 0.0, -9.0) # -9.0 is base value
|
||||
|
||||
#WORKING neuroplasticity
|
||||
wRec_cpu = neuroplasticity(synapticConnectionNumber, zit_cumulative_cpu, wRec_cpu,
|
||||
inactivityCounter_cpu)
|
||||
error("DEBUG -> lifLearn! $(Dates.now())")
|
||||
# #TODO send to device with correct dimension
|
||||
# wRec = wRec |> device
|
||||
# inactivityCounter = inactivityCounter_cpu |> device
|
||||
|
||||
end
|
||||
|
||||
function alifLearn!(wRec,
|
||||
wRecChange,
|
||||
arrayProjection4d,
|
||||
inactivityCounter,
|
||||
synapticConnectionNumber,
|
||||
zit_cumulative,
|
||||
device)
|
||||
# merge learning weight with average learning weight
|
||||
wRec .+= (sum(wRecChange, dims=4) ./ (size(wRec, 4))) .* arrayProjection4d
|
||||
|
||||
# weak / negative synaptic connection will get randomed in neuroplasticity()
|
||||
wRec .= GeneralUtils.replaceLessThan.(wRec, 0.01, 0.0)
|
||||
|
||||
#TODO synaptic strength
|
||||
|
||||
#TODO neuroplasticity
|
||||
|
||||
end
|
||||
|
||||
function onLearn!(wOut,
|
||||
wOutChange,
|
||||
arrayProjection4d)
|
||||
# merge learning weight with average learning weight
|
||||
wOut .+= (sum(wOutChange, dims=4) ./ (size(wOut, 4))) .* arrayProjection4d
|
||||
|
||||
# adaptive wOut to help convergence using c_decay
|
||||
wOut .-= 0.001 .* wOut
|
||||
|
||||
#TODO synaptic strength
|
||||
|
||||
#TODO neuroplasticity
|
||||
|
||||
end
|
||||
|
||||
function neuroplasticity(synapticConnectionNumber,
|
||||
zit_cumulative, # (row, col)
|
||||
wRec, # (row, col, n)
|
||||
inactivityCounter_cpu) # (row, col, n)
|
||||
|
||||
i1,i2,i3 = size(wRec)
|
||||
|
||||
# for each neuron, find total number of synaptic conn that should draw
|
||||
# new connection to firing and non-firing neurons pool
|
||||
subToFireNeuron_toBe = Int(floor(0.7 * synapticConnectionNumber))
|
||||
subToNonFiringNeuron_toBe = synapticConnectionNumber - subToFireNeuron_toBe
|
||||
|
||||
#WORKING for each neuron, count how many synap already subscribed to firing-neurons
|
||||
subToFireNeuron_current = sum((!iszero).(zit_cumulative .* wRec), dims=(1,2)) # (1, 1, n)
|
||||
subToNonFiringNeuron_current = synapticConnectionNumber .- subToFireNeuron_current # (1, 1, n)
|
||||
mask = (!iszero).(zit_cumulative) # mask of firing neurons = 1, non-firing = 0
|
||||
projection = ones(i1,i2,i3)
|
||||
mask = mask .* projection # (row, col, n)
|
||||
totalNewConn = sum(isequal.(wRec, -1.0), dims=(1,2)) # count new conn mark (-1.0), (1, 1, n)
|
||||
println("mask ", size(mask))
|
||||
println("wRec ", size(wRec))
|
||||
println("inactivityCounter_cpu ", size(inactivityCounter_cpu))
|
||||
println("totalNeurons ", totalNewConn, size(totalNewConn))
|
||||
error("DEBUG -> neuroplasticity $(Dates.now())")
|
||||
for i in 1:i3
|
||||
|
||||
|
||||
# add new conn to firing neurons pool
|
||||
remaining = GeneralUtils.replace_elements(mask[:,:,i],
|
||||
1,
|
||||
wRecmask[:,:,i],
|
||||
inactivityCounter_cpumask[:,:,i],
|
||||
totalNewConn[:,:,i])
|
||||
|
||||
#TODO add new conn to non-firing neurons pool
|
||||
|
||||
end
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
newFiringConn = subToFireNeuron_toBe - subToFireNeuron_current
|
||||
newFiringConn = newFiringConn > 0 ? newFiringConn : 0
|
||||
|
||||
newNonFiringConn = subToNonFiringNeuron_toBe - subToNonFiringNeuron_current
|
||||
|
||||
return wRec
|
||||
end
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
end # module
|
||||
87
previousVersion/0.0.05-alpha/src/snnUtil.jl
Normal file
87
previousVersion/0.0.05-alpha/src/snnUtil.jl
Normal file
@@ -0,0 +1,87 @@
|
||||
module snnUtil
|
||||
|
||||
export refractoryStatus!
|
||||
|
||||
# using
|
||||
|
||||
#------------------------------------------------------------------------------------------------100
|
||||
|
||||
function refractoryStatus!(refractoryCounter, refractoryActive, refractoryInactive)
|
||||
d1, d2, d3, d4 = size(refractoryCounter)
|
||||
for j in 1:d4
|
||||
for i in 1:d3
|
||||
if refractoryCounter[1, 1, i, j] > 0 # inactive
|
||||
view(refractoryActive, 1, 1, i, j) .= 0
|
||||
view(refractoryInactive, 1, 1, i, j) .= 1
|
||||
else # active
|
||||
view(refractoryActive, 1, 1, i, j) .= 1
|
||||
view(refractoryInactive, 1, 1, i, j) .= 0
|
||||
end
|
||||
end
|
||||
end
|
||||
end
|
||||
|
||||
function frobenius_distance(A, B)
|
||||
# Check if the matrices have the same size
|
||||
if size(A) != size(B)
|
||||
error("The matrices must have the same size")
|
||||
end
|
||||
# Initialize the distance to zero
|
||||
distance = 0.0
|
||||
# Loop over the elements of the matrices and add the squared differences
|
||||
for i in 1:size(A, 1)
|
||||
for j in 1:size(A, 2)
|
||||
distance += (A[i, j] - B[i, j])^2
|
||||
end
|
||||
end
|
||||
# Return the square root of the distance
|
||||
return sqrt(distance)
|
||||
end
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
end # module
|
||||
437
previousVersion/0.0.05-alpha/src/type.jl
Normal file
437
previousVersion/0.0.05-alpha/src/type.jl
Normal file
@@ -0,0 +1,437 @@
|
||||
module type
|
||||
|
||||
export
|
||||
# struct
|
||||
kfn_1
|
||||
|
||||
# function
|
||||
|
||||
using Random, GeneralUtils
|
||||
|
||||
#------------------------------------------------------------------------------------------------100
|
||||
rng = MersenneTwister(1234)
|
||||
|
||||
abstract type Ironpen end
|
||||
abstract type knowledgeFn <: Ironpen end
|
||||
|
||||
#------------------------------------------------------------------------------------------------100
|
||||
|
||||
Base.@kwdef mutable struct kfn_1 <: knowledgeFn
|
||||
params::Union{Dict, Nothing} = nothing # store params of knowledgeFn itself for later use
|
||||
|
||||
timeStep::Union{AbstractArray, Nothing} = nothing
|
||||
learningStage::Union{AbstractArray, Nothing} = nothing # 0 inference, 1 start, 2 during, 3 end learning
|
||||
inputSize::Union{AbstractArray, Nothing} = nothing
|
||||
zit::Union{AbstractArray, Nothing} = nothing # 3D activation matrix
|
||||
zit_cumulative::Union{AbstractArray, Nothing} = nothing
|
||||
exInType::Union{AbstractArray, Nothing} = nothing
|
||||
modelError::Union{AbstractArray, Nothing} = nothing # store RSNN error
|
||||
outputError::Union{AbstractArray, Nothing} = nothing # store output neurons error
|
||||
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# LIF Neurons #
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# a projection of kfn.zit into lif dimension for broadcasting later)
|
||||
lif_zit::Union{AbstractArray, Nothing} = nothing
|
||||
|
||||
# main variables according to papers
|
||||
lif_wRec::Union{AbstractArray, Nothing} = nothing
|
||||
lif_vt::Union{AbstractArray, Nothing} = nothing
|
||||
lif_vth::Union{AbstractArray, Nothing} = nothing
|
||||
lif_vRest::Union{AbstractArray, Nothing} = nothing
|
||||
lif_zt::Union{AbstractArray, Nothing} = nothing
|
||||
lif_zt4d::Union{AbstractArray, Nothing} = nothing
|
||||
lif_refractoryCounter::Union{AbstractArray, Nothing} = nothing
|
||||
lif_refractoryDuration::Union{AbstractArray, Nothing} = nothing
|
||||
lif_alpha::Union{AbstractArray, Nothing} = nothing
|
||||
lif_delta::Union{AbstractFloat, Nothing} = nothing
|
||||
lif_tau_m::Union{AbstractFloat, Nothing} = nothing
|
||||
lif_phi::Union{AbstractArray, Nothing} = nothing
|
||||
lif_epsilonRec::Union{AbstractArray, Nothing} = nothing
|
||||
lif_eRec::Union{AbstractArray, Nothing} = nothing
|
||||
lif_eta::Union{AbstractArray, Nothing} = nothing
|
||||
lif_gammaPd::Union{AbstractArray, Nothing} = nothing
|
||||
lif_wRecChange::Union{AbstractArray, Nothing} = nothing
|
||||
lif_error::Union{AbstractArray, Nothing} = nothing
|
||||
|
||||
lif_firingCounter::Union{AbstractArray, Nothing} = nothing
|
||||
lif_firingTargetFrequency::Union{AbstractArray, Nothing} = nothing
|
||||
lif_neuronInactivityCounter::Union{AbstractArray, Nothing} = nothing
|
||||
lif_synapticInactivityCounter::Union{AbstractArray, Nothing} = nothing
|
||||
lif_synapticConnectionNumber::Union{Int, Nothing} = nothing
|
||||
|
||||
# pre-allocation array
|
||||
lif_arrayProjection4d::Union{AbstractArray, Nothing} = nothing # use to project 3d array to 4d
|
||||
lif_recSignal::Union{AbstractArray, Nothing} = nothing
|
||||
lif_exInType::Union{AbstractArray, Nothing} = nothing
|
||||
# lif_decayed_epsilonRec::Union{AbstractArray, Nothing} = nothing
|
||||
# lif_vt_diff_vth::Union{AbstractArray, Nothing} = nothing
|
||||
# lif_vt_diff_vth_div_vth::Union{AbstractArray, Nothing} = nothing
|
||||
# lif_gammaPd_div_vth::Union{AbstractArray, Nothing} = nothing
|
||||
# lif_phiActivation::Union{AbstractArray, Nothing} = nothing
|
||||
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# ALIF Neurons #
|
||||
# ---------------------------------------------------------------------------- #
|
||||
alif_zit::Union{AbstractArray, Nothing} = nothing
|
||||
|
||||
alif_wRec::Union{AbstractArray, Nothing} = nothing
|
||||
alif_vt::Union{AbstractArray, Nothing} = nothing
|
||||
alif_vth::Union{AbstractArray, Nothing} = nothing
|
||||
alif_vRest::Union{AbstractArray, Nothing} = nothing
|
||||
alif_zt::Union{AbstractArray, Nothing} = nothing
|
||||
alif_zt4d::Union{AbstractArray, Nothing} = nothing
|
||||
alif_refractoryCounter::Union{AbstractArray, Nothing} = nothing
|
||||
alif_refractoryDuration::Union{AbstractArray, Nothing} = nothing
|
||||
alif_alpha::Union{AbstractArray, Nothing} = nothing
|
||||
alif_delta::Union{AbstractFloat, Nothing} = nothing
|
||||
alif_tau_m::Union{AbstractFloat, Nothing} = nothing
|
||||
alif_phi::Union{AbstractArray, Nothing} = nothing
|
||||
alif_epsilonRec::Union{AbstractArray, Nothing} = nothing
|
||||
alif_eRec::Union{AbstractArray, Nothing} = nothing
|
||||
alif_eta::Union{AbstractArray, Nothing} = nothing
|
||||
alif_gammaPd::Union{AbstractArray, Nothing} = nothing
|
||||
alif_wRecChange::Union{AbstractArray, Nothing} = nothing
|
||||
alif_error::Union{AbstractArray, Nothing} = nothing
|
||||
|
||||
alif_firingCounter::Union{AbstractArray, Nothing} = nothing
|
||||
alif_firingTargetFrequency::Union{AbstractArray, Nothing} = nothing
|
||||
alif_neuronInactivityCounter::Union{AbstractArray, Nothing} = nothing
|
||||
alif_synapticInactivityCounter::Union{AbstractArray, Nothing} = nothing
|
||||
alif_synapticConnectionNumber::Union{Int, Nothing} = nothing
|
||||
|
||||
# pre-allocation array
|
||||
alif_arrayProjection4d::Union{AbstractArray, Nothing} = nothing # use to project 3d array to 4d
|
||||
alif_recSignal::Union{AbstractArray, Nothing} = nothing
|
||||
alif_exInType::Union{AbstractArray, Nothing} = nothing
|
||||
# alif_decayed_epsilonRec::Union{AbstractArray, Nothing} = nothing
|
||||
# alif_vt_diff_vth::Union{AbstractArray, Nothing} = nothing
|
||||
# alif_vt_diff_vth_div_vth::Union{AbstractArray, Nothing} = nothing
|
||||
# alif_gammaPd_div_vth::Union{AbstractArray, Nothing} = nothing
|
||||
# alif_phiActivation::Union{AbstractArray, Nothing} = nothing
|
||||
|
||||
# alif specific variables
|
||||
alif_epsilonRecA::Union{AbstractArray, Nothing} = nothing
|
||||
alif_avth::Union{AbstractArray, Nothing} = nothing
|
||||
alif_a::Union{AbstractArray, Nothing} = nothing # threshold adaptation
|
||||
alif_beta::Union{AbstractArray, Nothing} = nothing # β, constant, value from paper
|
||||
alif_rho::Union{AbstractArray, Nothing} = nothing # ρ, threshold adaptation decay factor
|
||||
alif_tau_a::Union{AbstractFloat, Nothing} = nothing # τ_a, adaption time constant in millisecond
|
||||
|
||||
# alif specific pre-allocation array
|
||||
# alif_phi_x_epsilonRec::Union{AbstractArray, Nothing} = nothing
|
||||
# alif_phi_x_beta::Union{AbstractArray, Nothing} = nothing
|
||||
# alif_rho_diff_phi_x_beta::Union{AbstractArray, Nothing} = nothing
|
||||
# alif_rho_div_phi_x_beta_x_epsilonRecA::Union{AbstractArray, Nothing} = nothing
|
||||
# alif_beta_x_a::Union{AbstractArray, Nothing} = nothing
|
||||
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# Output Neurons #
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# output neuron is based on LIF
|
||||
on_zit::Union{AbstractArray, Nothing} = nothing
|
||||
|
||||
# main variables according to papers
|
||||
on_wOut::Union{AbstractArray, Nothing} = nothing # wOut is wRec, just use the name from paper
|
||||
on_vt::Union{AbstractArray, Nothing} = nothing
|
||||
on_vth::Union{AbstractArray, Nothing} = nothing
|
||||
on_vRest::Union{AbstractArray, Nothing} = nothing
|
||||
on_zt::Union{AbstractArray, Nothing} = nothing
|
||||
on_zt4d::Union{AbstractArray, Nothing} = nothing
|
||||
on_refractoryCounter::Union{AbstractArray, Nothing} = nothing
|
||||
on_refractoryDuration::Union{AbstractArray, Nothing} = nothing
|
||||
on_alpha::Union{AbstractArray, Nothing} = nothing
|
||||
on_delta::Union{AbstractFloat, Nothing} = nothing
|
||||
on_tau_m::Union{AbstractFloat, Nothing} = nothing
|
||||
on_phi::Union{AbstractArray, Nothing} = nothing
|
||||
on_epsilonRec::Union{AbstractArray, Nothing} = nothing
|
||||
on_eRec::Union{AbstractArray, Nothing} = nothing
|
||||
on_eta::Union{AbstractArray, Nothing} = nothing
|
||||
on_gammaPd::Union{AbstractArray, Nothing} = nothing
|
||||
on_wOutChange::Union{AbstractArray, Nothing} = nothing
|
||||
on_error::Union{AbstractArray, Nothing} = nothing
|
||||
on_subscription::Union{AbstractArray, Nothing} = nothing
|
||||
|
||||
on_firingCounter::Union{AbstractArray, Nothing} = nothing
|
||||
|
||||
# pre-allocation array
|
||||
on_arrayProjection4d::Union{AbstractArray, Nothing} = nothing # use to project 3d array to 4d
|
||||
on_recSignal::Union{AbstractArray, Nothing} = nothing
|
||||
# on_decayed_epsilonRec::Union{AbstractArray, Nothing} = nothing
|
||||
# on_vt_diff_vth::Union{AbstractArray, Nothing} = nothing
|
||||
# on_vt_diff_vth_div_vth::Union{AbstractArray, Nothing} = nothing
|
||||
# on_gammaPd_div_vth::Union{AbstractArray, Nothing} = nothing
|
||||
# on_phiActivation::Union{AbstractArray, Nothing} = nothing
|
||||
end
|
||||
|
||||
# outer constructor
|
||||
function kfn_1(params::Dict; device=cpu)
|
||||
kfn = kfn_1()
|
||||
kfn.params = params
|
||||
kfn.timeStep = [0] |> device
|
||||
kfn.learningStage = [0] |> device
|
||||
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# initialize activation matrix #
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# row*col is a 2D matrix represent all RSNN activation
|
||||
row, signal_col, batch = kfn.params[:inputPort][:signal][:numbers] # z-axis represent signal batch number
|
||||
|
||||
kfn.inputSize = [row, signal_col] |> device
|
||||
lif_col = kfn.params[:computeNeuron][:lif][:numbers][2]
|
||||
alif_col = kfn.params[:computeNeuron][:alif][:numbers][2]
|
||||
|
||||
col = signal_col + lif_col + alif_col
|
||||
|
||||
# activation matrix
|
||||
kfn.zit = zeros(row, col, batch) |> device
|
||||
kfn.zit_cumulative = (similar(kfn.zit) .= 0)
|
||||
kfn.modelError = zeros(1) |> device
|
||||
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# LIF config #
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# In 3D LIF matrix, z-axis represent each neuron while each 2D slice represent that neuron's
|
||||
# synaptic subscription to other neurons (via activation matrix)
|
||||
lif_n = kfn.params[:computeNeuron][:lif][:numbers][1] * kfn.params[:computeNeuron][:lif][:numbers][2]
|
||||
|
||||
# subscription
|
||||
w = zeros(row, col, lif_n)
|
||||
synapticConnectionPercent = kfn.params[:computeNeuron][:lif][:params][:synapticConnectionPercent]
|
||||
kfn.lif_synapticConnectionNumber = Int(floor(row*col * synapticConnectionPercent/100))
|
||||
for slice in eachslice(w, dims=3)
|
||||
pool = shuffle!([1:row*col...])[1:kfn.lif_synapticConnectionNumber]
|
||||
for i in pool
|
||||
slice[i] = rand() # assign weight to synaptic connection. /10 to start small,
|
||||
# otherwise RSNN's vt Usually stay negative (-)
|
||||
end
|
||||
end
|
||||
|
||||
# 10% of neuron connection should be enough to start to make neuron fires
|
||||
should_be_avg_weight = 1 / (0.1 * lif_n)
|
||||
w = w .* (should_be_avg_weight / maximum(w)) # adjust overall weight
|
||||
|
||||
# project 3D w into 4D kfn.lif_wRec (row, col, n, batch)
|
||||
kfn.lif_wRec = reshape(w, (row, col, lif_n, 1)) .* ones(row, col, lif_n, batch) |> device
|
||||
kfn.lif_zit = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_vt = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_vth = (similar(kfn.lif_wRec) .= 1)
|
||||
kfn.lif_vRest = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_zt = zeros(1, 1, lif_n, batch) |> device
|
||||
kfn.lif_zt4d = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_refractoryCounter = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_refractoryDuration = (similar(kfn.lif_wRec) .= 3)
|
||||
kfn.lif_delta = 1.0
|
||||
kfn.lif_tau_m = 20.0
|
||||
kfn.lif_alpha = (similar(kfn.lif_wRec) .= (exp(-kfn.lif_delta / kfn.lif_tau_m)))
|
||||
kfn.lif_phi = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_epsilonRec = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_eRec = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_eta = (similar(kfn.lif_wRec) .= 0.001)
|
||||
kfn.lif_gammaPd = (similar(kfn.lif_wRec) .= 0.3)
|
||||
kfn.lif_wRecChange = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_error = (similar(kfn.lif_wRec) .= 0)
|
||||
|
||||
kfn.lif_firingCounter = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_firingTargetFrequency = (similar(kfn.lif_wRec) .= 0.1)
|
||||
kfn.lif_neuronInactivityCounter = (similar(kfn.lif_wRec) .= 10000)
|
||||
kfn.lif_synapticInactivityCounter = Array(similar(kfn.lif_wRec) .= -9) # -9 for non-sub conn
|
||||
mask = Array((!iszero).(kfn.lif_wRec))
|
||||
GeneralUtils.replace_elements!(mask, 1, kfn.lif_synapticInactivityCounter, 10000)
|
||||
kfn.lif_synapticInactivityCounter = kfn.lif_synapticInactivityCounter |> device
|
||||
|
||||
kfn.lif_arrayProjection4d = (similar(kfn.lif_wRec) .= 1)
|
||||
kfn.lif_recSignal = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_exInType = (similar(kfn.lif_wRec) .= 0)
|
||||
# kfn.lif_decayed_epsilonRec = (similar(kfn.lif_wRec) .= 0)
|
||||
# kfn.lif_vt_diff_vth = (similar(kfn.lif_wRec) .= 0)
|
||||
# kfn.lif_vt_diff_vth_div_vth = (similar(kfn.lif_wRec) .= 0)
|
||||
# kfn.lif_gammaPd_div_vth = (similar(kfn.lif_wRec) .= 0)
|
||||
# kfn.lif_phiActivation = (similar(kfn.lif_wRec) .= 0)
|
||||
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# ALIF config #
|
||||
# ---------------------------------------------------------------------------- #
|
||||
alif_n = kfn.params[:computeNeuron][:alif][:numbers][1] * kfn.params[:computeNeuron][:alif][:numbers][2]
|
||||
|
||||
# subscription
|
||||
w = zeros(row, col, alif_n)
|
||||
synapticConnectionPercent = kfn.params[:computeNeuron][:alif][:params][:synapticConnectionPercent]
|
||||
kfn.alif_synapticConnectionNumber = Int(floor(row*col * synapticConnectionPercent/100))
|
||||
for slice in eachslice(w, dims=3)
|
||||
pool = shuffle!([1:row*col...])[1:kfn.alif_synapticConnectionNumber]
|
||||
for i in pool
|
||||
slice[i] = rand() # assign weight to synaptic connection. /10 to start small,
|
||||
# otherwise RSNN's vt Usually stay negative (-)
|
||||
end
|
||||
end
|
||||
|
||||
# 10% of neuron connection should be enough to start to make neuron fires
|
||||
should_be_avg_weight = 1 / (0.1 * alif_n)
|
||||
w = w .* (should_be_avg_weight / maximum(w)) # adjust overall weight
|
||||
|
||||
# project 3D w into 4D kfn.alif_wRec
|
||||
kfn.alif_wRec = reshape(w, (row, col, alif_n, 1)) .* ones(row, col, alif_n, batch) |> device
|
||||
kfn.alif_zit = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_vt = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_vth = (similar(kfn.alif_wRec) .= 1)
|
||||
kfn.alif_vRest = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_zt = zeros(1, 1, alif_n, batch) |> device
|
||||
kfn.alif_zt4d = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_refractoryCounter = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_refractoryDuration = (similar(kfn.alif_wRec) .= 3)
|
||||
kfn.alif_delta = 1.0
|
||||
kfn.alif_tau_m = 20.0
|
||||
kfn.alif_alpha = (similar(kfn.alif_wRec) .= (exp(-kfn.alif_delta / kfn.alif_tau_m)))
|
||||
kfn.alif_phi = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_epsilonRec = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_eRec = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_eta = (similar(kfn.alif_wRec) .= 0.001)
|
||||
kfn.alif_gammaPd = (similar(kfn.alif_wRec) .= 0.3)
|
||||
kfn.alif_wRecChange = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_error = (similar(kfn.alif_wRec) .= 0)
|
||||
|
||||
kfn.alif_firingCounter = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_firingTargetFrequency = (similar(kfn.alif_wRec) .= 0.1)
|
||||
kfn.alif_neuronInactivityCounter = (similar(kfn.alif_wRec) .= 10000)
|
||||
kfn.alif_synapticInactivityCounter = Array(similar(kfn.alif_wRec) .= -9) # -9 for non-sub conn
|
||||
mask = Array((!iszero).(kfn.alif_wRec))
|
||||
GeneralUtils.replace_elements!(mask, 1, kfn.alif_synapticInactivityCounter, 10000)
|
||||
kfn.alif_synapticInactivityCounter = kfn.alif_synapticInactivityCounter |> device
|
||||
|
||||
kfn.alif_arrayProjection4d = (similar(kfn.alif_wRec) .= 1)
|
||||
kfn.alif_recSignal = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_exInType = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_decayed_epsilonRec = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_vt_diff_vth = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_vt_diff_vth_div_vth = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_gammaPd_div_vth = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_phiActivation = (similar(kfn.alif_wRec) .= 0)
|
||||
|
||||
# alif specific variables
|
||||
kfn.alif_epsilonRecA = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_avth = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_a = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_beta = (similar(kfn.alif_wRec) .= 0.07)
|
||||
kfn.alif_tau_a = 800.0
|
||||
kfn.alif_rho = (similar(kfn.alif_wRec) .= (exp(-kfn.alif_delta / kfn.alif_tau_a))) |> device
|
||||
# kfn.alif_phi_x_epsilonRec = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_phi_x_beta = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_rho_diff_phi_x_beta = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_rho_div_phi_x_beta_x_epsilonRecA = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_beta_x_a = (similar(kfn.alif_wRec) .= 0)
|
||||
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# output config #
|
||||
# ---------------------------------------------------------------------------- #
|
||||
n = kfn.params[:outputPort][:numbers][1] * kfn.params[:outputPort][:numbers][2]
|
||||
|
||||
# subscription
|
||||
w = zeros(row, col, n)
|
||||
synapticConnectionPercent = kfn.params[:outputPort][:params][:synapticConnectionPercent]
|
||||
subable = size(kfn.lif_wRec, 3) + size(kfn.alif_wRec, 3) # sub to lif, alif only
|
||||
synapticConnection = Int(floor(subable * synapticConnectionPercent/100))
|
||||
for slice in eachslice(w, dims=3) # each slice is a neuron
|
||||
startInd = row*col - subable + 1 # e.g. 100(row*col) - 50(subable) = 50 -> startInd = 51
|
||||
|
||||
# pool must contain only lif, alif neurons
|
||||
pool = shuffle!([startInd:row*col...])[1:synapticConnection]
|
||||
for i in pool
|
||||
slice[i] = rand() # assign weight to synaptic connection. /10 to start small,
|
||||
# otherwise RSNN's vt Usually stay negative (-)
|
||||
end
|
||||
end
|
||||
|
||||
# # 10% of neuron connection should be enough to start to make neuron fires
|
||||
# should_be_avg_weight = 1 / (0.2 * n)
|
||||
# w = w .* (should_be_avg_weight / maximum(w)) # adjust overall weight
|
||||
|
||||
# project 3D w into 4D kfn.lif_wOut (row, col, n, batch)
|
||||
kfn.on_wOut = reshape(w, (row, col, n, 1)) .* ones(row, col, n, batch) |> device
|
||||
kfn.on_zit = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_vt = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_vth = (similar(kfn.on_wOut) .= 1)
|
||||
kfn.on_vRest = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_zt = zeros(1, 1, n, batch) |> device
|
||||
kfn.on_zt4d = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_refractoryCounter = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_refractoryDuration = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_delta = 1.0
|
||||
kfn.on_tau_m = 20.0
|
||||
kfn.on_alpha = (similar(kfn.on_wOut) .= (exp(-kfn.on_delta / kfn.on_tau_m)))
|
||||
kfn.on_phi = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_epsilonRec = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_eRec = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_eta = (similar(kfn.on_wOut) .= 0.001)
|
||||
kfn.on_gammaPd = (similar(kfn.on_wOut) .= 0.3)
|
||||
kfn.on_wOutChange = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_error = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_subscription = (GeneralUtils.isNotEqual.(kfn.on_wOut, 0)) |> device
|
||||
|
||||
kfn.on_firingCounter = (similar(kfn.on_wOut) .= 0)
|
||||
|
||||
kfn.on_arrayProjection4d = (similar(kfn.on_wOut) .= 1)
|
||||
kfn.on_recSignal = (similar(kfn.on_wOut) .= 0)
|
||||
|
||||
kfn.outputError = zeros(n, batch) |> device
|
||||
totalComputeNeurons = lif_n + alif_n
|
||||
inhabitoryNeurons = Int(floor(totalComputeNeurons * 30/100))
|
||||
mask1 = ones(row, signal_col)
|
||||
mask2 = GeneralUtils.multiply_random_elements(ones(row, lif_col + alif_col),
|
||||
-1, inhabitoryNeurons, MersenneTwister(1234))
|
||||
kfn.exInType = cat(mask1, mask2, dims=2) |> device
|
||||
|
||||
return kfn
|
||||
end
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
end # module
|
||||
@@ -25,19 +25,22 @@ using .interface
|
||||
|
||||
#------------------------------------------------------------------------------------------------100
|
||||
|
||||
""" version 0.0.3
|
||||
""" version 0.0.5
|
||||
Todo:
|
||||
[2] implement dormant connection and pruning machanism. the longer the training the longer
|
||||
[DONE] add excitatory/inhabitory matrix
|
||||
[-] add temporal summation in addition to already used spatial summation.
|
||||
CANCELLED, spatial summation every second until membrane potential reach a threshold
|
||||
is in itself a temporal summation.
|
||||
[x] add neuroplasticity
|
||||
[4] implement dormant connection and pruning machanism. the longer the training the longer
|
||||
0 weight stay 0.
|
||||
[] using RL to control learning signal
|
||||
[] consider using Dates.now() instead of timestamp because time_stamp may overflow
|
||||
[] Liquid time constant. training should include adjusting α, neuron membrane potential decay factor
|
||||
which defined by neuron.tau_m formula in type.jl
|
||||
|
||||
Change from version: 0.0.2
|
||||
- knowledgeFn in GPU format
|
||||
- use partial error update for computeNeuron
|
||||
- frequency regulator
|
||||
Change from version: 0.0.4
|
||||
-
|
||||
|
||||
All features
|
||||
|
||||
|
||||
@@ -55,6 +55,7 @@ function (kfn::kfn_1)(input::AbstractArray)
|
||||
# project 3D kfn zit into 4D lif zit
|
||||
i1, i2, i3, i4 = size(kfn.lif_zit)
|
||||
kfn.lif_zit .= reshape(kfn.zit, (i1, i2, 1, i4)) .* kfn.lif_arrayProjection4d
|
||||
kfn.lif_exInType .= kfn.exInType .* kfn.lif_arrayProjection4d
|
||||
|
||||
lifForward( kfn.lif_zit,
|
||||
kfn.lif_wRec,
|
||||
@@ -70,13 +71,16 @@ function (kfn::kfn_1)(input::AbstractArray)
|
||||
kfn.lif_gammaPd,
|
||||
kfn.lif_firingCounter,
|
||||
kfn.lif_recSignal,
|
||||
kfn.lif_subscription,
|
||||
kfn.lif_exInType,
|
||||
kfn.lif_neuronInactivityCounter,
|
||||
kfn.lif_synapticInactivityCounter,
|
||||
)
|
||||
end
|
||||
@async begin
|
||||
# project 3D kfn zit into 4D alif zit
|
||||
i1, i2, i3, i4 = size(kfn.alif_zit)
|
||||
kfn.alif_zit .= reshape(kfn.zit, (i1, i2, 1, i4)) .* kfn.alif_arrayProjection4d
|
||||
kfn.alif_exInType .= kfn.exInType .* kfn.alif_arrayProjection4d
|
||||
|
||||
alifForward(kfn.alif_zit,
|
||||
kfn.alif_wRec,
|
||||
@@ -92,12 +96,15 @@ function (kfn::kfn_1)(input::AbstractArray)
|
||||
kfn.alif_gammaPd,
|
||||
kfn.alif_firingCounter,
|
||||
kfn.alif_recSignal,
|
||||
kfn.alif_subscription,
|
||||
kfn.alif_exInType,
|
||||
kfn.alif_neuronInactivityCounter,
|
||||
kfn.alif_synapticInactivityCounter,
|
||||
kfn.alif_epsilonRecA,
|
||||
kfn.alif_a,
|
||||
kfn.alif_avth,
|
||||
kfn.alif_beta,
|
||||
kfn.alif_rho,)
|
||||
kfn.alif_rho,
|
||||
)
|
||||
end
|
||||
end
|
||||
|
||||
@@ -132,13 +139,11 @@ function (kfn::kfn_1)(input::AbstractArray)
|
||||
kfn.on_gammaPd,
|
||||
kfn.on_firingCounter,
|
||||
kfn.on_recSignal,
|
||||
kfn.on_subscription,
|
||||
)
|
||||
# get on_zt4d to on_zt
|
||||
kfn.on_zt .= reduce(max, kfn.on_zt4d, dims=(1,2))
|
||||
logit = reshape(kfn.on_zt, (size(input, 1), :))
|
||||
|
||||
# error("DEBUG -> kfn forward")
|
||||
return logit,
|
||||
kfn.zit
|
||||
end
|
||||
@@ -158,7 +163,9 @@ function lifForward( zit::CuArray,
|
||||
gammaPd::CuArray,
|
||||
firingCounter::CuArray,
|
||||
recSignal::CuArray,
|
||||
subscription::CuArray,
|
||||
exInType::CuArray,
|
||||
neuronInactivityCounter::CuArray,
|
||||
synapticInactivityCounter::CuArray,
|
||||
)
|
||||
|
||||
kernel = @cuda launch=false lifForward( zit,
|
||||
@@ -175,7 +182,9 @@ function lifForward( zit::CuArray,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
subscription,
|
||||
exInType,
|
||||
neuronInactivityCounter,
|
||||
synapticInactivityCounter,
|
||||
GeneralUtils.linear_to_cartesian,
|
||||
)
|
||||
config = launch_configuration(kernel.fun)
|
||||
@@ -206,7 +215,9 @@ function lifForward( zit::CuArray,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
subscription,
|
||||
exInType,
|
||||
neuronInactivityCounter,
|
||||
synapticInactivityCounter,
|
||||
GeneralUtils.linear_to_cartesian; threads, blocks)
|
||||
end
|
||||
end
|
||||
@@ -226,7 +237,9 @@ function lifForward( zit,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
subscription,
|
||||
exInType,
|
||||
neuronInactivityCounter,
|
||||
synapticInactivityCounter,
|
||||
linear_to_cartesian,
|
||||
)
|
||||
i = (blockIdx().x - 1) * blockDim().x + threadIdx().x # gpu threads index
|
||||
@@ -247,7 +260,8 @@ function lifForward( zit,
|
||||
epsilonRec[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * epsilonRec[i1,i2,i3,i4])
|
||||
|
||||
else # refractory period is inactive
|
||||
recSignal[i1,i2,i3,i4] = wRec[i1,i2,i3,i4] * zit[i1,i2,i3,i4]
|
||||
recSignal[i1,i2,i3,i4] = wRec[i1,i2,i3,i4] * zit[i1,i2,i3,i4] *
|
||||
exInType[i1,i2,i3,i4]
|
||||
vt[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * vt[i1,i2,i3,i4]) +
|
||||
sum(@view(recSignal[:,:,i3,i4]))
|
||||
|
||||
@@ -257,8 +271,12 @@ function lifForward( zit,
|
||||
refractoryCounter[i1,i2,i3,i4] = refractoryDuration[i1,i2,i3,i4]
|
||||
firingCounter[i1,i2,i3,i4] += 1
|
||||
vt[i1,i2,i3,i4] = vRest[i1,i2,i3,i4]
|
||||
|
||||
# reset counter if neuron fires
|
||||
neuronInactivityCounter[i1,i2,i3,i4] = 10000
|
||||
else
|
||||
zt[i1,i2,i3,i4] = 0
|
||||
neuronInactivityCounter[i1,i2,i3,i4] -= 1
|
||||
end
|
||||
|
||||
# compute phi, there is a difference from lif formula
|
||||
@@ -267,7 +285,18 @@ function lifForward( zit,
|
||||
|
||||
# compute epsilonRec
|
||||
epsilonRec[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * epsilonRec[i1,i2,i3,i4]) +
|
||||
(zit[i1,i2,i3,i4] * subscription[i1,i2,i3,i4])
|
||||
(zit[i1,i2,i3,i4] * !iszero(wRec[i1,i2,i3,i4]))
|
||||
# !iszero indicates synaptic subscription
|
||||
|
||||
# count synaptic inactivity
|
||||
if !iszero(wRec[i1,i2,i3,i4]) # check if this is wRec subscription
|
||||
if !iszero(zit[i1,i2,i3,i4]) # synapse is active, reset counter
|
||||
synapticInactivityCounter[i1,i2,i3,i4] = 10000
|
||||
else # synapse is inactive, counting
|
||||
synapticInactivityCounter[i1,i2,i3,i4] -= 1
|
||||
end
|
||||
end
|
||||
|
||||
end
|
||||
end
|
||||
return nothing
|
||||
@@ -288,7 +317,9 @@ function alifForward( zit::CuArray,
|
||||
gammaPd::CuArray,
|
||||
firingCounter::CuArray,
|
||||
recSignal::CuArray,
|
||||
subscription::CuArray,
|
||||
exInType::CuArray,
|
||||
neuronInactivityCounter::CuArray,
|
||||
synapticInactivityCounter::CuArray,
|
||||
epsilonRecA::CuArray,
|
||||
a::CuArray,
|
||||
avth::CuArray,
|
||||
@@ -310,7 +341,9 @@ function alifForward( zit::CuArray,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
subscription,
|
||||
exInType,
|
||||
neuronInactivityCounter,
|
||||
synapticInactivityCounter,
|
||||
epsilonRecA,
|
||||
a,
|
||||
avth,
|
||||
@@ -345,7 +378,9 @@ function alifForward( zit::CuArray,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
subscription,
|
||||
exInType,
|
||||
neuronInactivityCounter,
|
||||
synapticInactivityCounter,
|
||||
epsilonRecA,
|
||||
a,
|
||||
avth,
|
||||
@@ -370,7 +405,9 @@ function alifForward( zit,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
subscription,
|
||||
exInType,
|
||||
neuronInactivityCounter,
|
||||
synapticInactivityCounter,
|
||||
epsilonRecA,
|
||||
a,
|
||||
avth,
|
||||
@@ -404,7 +441,8 @@ function alifForward( zit,
|
||||
avth[i1,i2,i3,i4] = vth[i1,i2,i3,i4] + (beta[i1,i2,i3,i4] * a[i1,i2,i3,i4])
|
||||
|
||||
else # refractory period is inactive
|
||||
recSignal[i1,i2,i3,i4] = zit[i1,i2,i3,i4] * wRec[i1,i2,i3,i4]
|
||||
recSignal[i1,i2,i3,i4] = wRec[i1,i2,i3,i4] * zit[i1,i2,i3,i4] *
|
||||
exInType[i1,i2,i3,i4]
|
||||
vt[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * vt[i1,i2,i3,i4]) +
|
||||
sum(@view(recSignal[:,:,i3,i4]))
|
||||
|
||||
@@ -418,9 +456,11 @@ function alifForward( zit,
|
||||
firingCounter[i1,i2,i3,i4] += 1
|
||||
vt[i1,i2,i3,i4] = vRest[i1,i2,i3,i4]
|
||||
a[i1,i2,i3,i4] = (rho[i1,i2,i3,i4] * a[i1,i2,i3,i4]) + 1
|
||||
neuronInactivityCounter[i1,i2,i3,i4] = 10000
|
||||
else
|
||||
zt[i1,i2,i3,i4] = 0
|
||||
a[i1,i2,i3,i4] = (rho[i1,i2,i3,i4] * a[i1,i2,i3,i4])
|
||||
neuronInactivityCounter[i1,i2,i3,i4] -= 1
|
||||
end
|
||||
|
||||
# compute phi, there is a difference from alif formula
|
||||
@@ -429,11 +469,20 @@ function alifForward( zit,
|
||||
|
||||
# compute epsilonRec
|
||||
epsilonRec[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * epsilonRec[i1,i2,i3,i4]) +
|
||||
(zit[i1,i2,i3,i4] * subscription[i1,i2,i3,i4])
|
||||
(zit[i1,i2,i3,i4] * !iszero(wRec[i1,i2,i3,i4]))
|
||||
# compute epsilonRecA use eq.26
|
||||
epsilonRecA[i1,i2,i3,i4] = (rho[i1,i2,i3,i4] *
|
||||
(phi[i1,i2,i3,i4] * epsilonRec[i1,i2,i3,i4])) +
|
||||
(zit[i1,i2,i3,i4] * subscription[i1,i2,i3,i4])
|
||||
(phi[i1,i2,i3,i4] * epsilonRec[i1,i2,i3,i4])) +
|
||||
(zit[i1,i2,i3,i4] * !iszero(wRec[i1,i2,i3,i4]))
|
||||
|
||||
# count synaptic inactivity
|
||||
if !iszero(wRec[i1,i2,i3,i4]) # check if this is wRec subscription
|
||||
if !iszero(zit[i1,i2,i3,i4]) # synapse is active, reset counter
|
||||
synapticInactivityCounter[i1,i2,i3,i4] = 10000
|
||||
else # synapse is inactive, counting
|
||||
synapticInactivityCounter[i1,i2,i3,i4] -= 1
|
||||
end
|
||||
end
|
||||
end
|
||||
end
|
||||
return nothing
|
||||
@@ -454,7 +503,6 @@ function onForward( zit::CuArray,
|
||||
gammaPd::CuArray,
|
||||
firingCounter::CuArray,
|
||||
recSignal::CuArray,
|
||||
subscription::CuArray,
|
||||
)
|
||||
|
||||
kernel = @cuda launch=false onForward( zit,
|
||||
@@ -471,7 +519,6 @@ function onForward( zit::CuArray,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
subscription,
|
||||
GeneralUtils.linear_to_cartesian,
|
||||
)
|
||||
config = launch_configuration(kernel.fun)
|
||||
@@ -501,7 +548,6 @@ function onForward( zit::CuArray,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
subscription,
|
||||
GeneralUtils.linear_to_cartesian; threads, blocks)
|
||||
end
|
||||
end
|
||||
@@ -521,7 +567,6 @@ function onForward( zit,
|
||||
gammaPd,
|
||||
firingCounter,
|
||||
recSignal,
|
||||
subscription,
|
||||
linear_to_cartesian,
|
||||
)
|
||||
i = (blockIdx().x - 1) * blockDim().x + threadIdx().x # gpu threads index
|
||||
@@ -556,11 +601,12 @@ function onForward( zit,
|
||||
end
|
||||
|
||||
# compute phi, there is a difference from on formula
|
||||
phi[i1,i2,i3,i4] = (gammaPd[i1,i2,i3,i4] / vth[i1,i2,i3,i4]) * max(0, 1 - ((vt[i1,i2,i3,i4] - vth[i1,i2,i3,i4]) / vth[i1,i2,i3,i4]))
|
||||
phi[i1,i2,i3,i4] = (gammaPd[i1,i2,i3,i4] / vth[i1,i2,i3,i4]) *
|
||||
max(0, 1 - ((vt[i1,i2,i3,i4] - vth[i1,i2,i3,i4]) / vth[i1,i2,i3,i4]))
|
||||
|
||||
# compute epsilonRec
|
||||
epsilonRec[i1,i2,i3,i4] = (alpha[i1,i2,i3,i4] * epsilonRec[i1,i2,i3,i4]) +
|
||||
(zit[i1,i2,i3,i4] * subscription[i1,i2,i3,i4])
|
||||
(zit[i1,i2,i3,i4] * !iszero(wOut[i1,i2,i3,i4]))
|
||||
end
|
||||
end
|
||||
return nothing
|
||||
|
||||
105
src/learn.jl
105
src/learn.jl
@@ -171,7 +171,7 @@ function onComputeParamsChange!(phi::CuArray,
|
||||
|
||||
eRec .= phi .* epsilonRec
|
||||
nError .= reshape(outputError, (1, 1, :, size(outputError, 2))) .* arrayProjection4d
|
||||
wOutChange .+= (-eta .* nError .* eRec) #BUG why wOutChange not increase every timestep that madel get wrong answer?
|
||||
wOutChange .+= (-eta .* nError .* eRec)
|
||||
|
||||
# reset epsilonRec
|
||||
epsilonRec .= 0
|
||||
@@ -264,16 +264,24 @@ function onComputeParamsChange!(phi::AbstractArray,
|
||||
end
|
||||
end
|
||||
|
||||
function learn!(kfn::kfn_1)
|
||||
function learn!(kfn::kfn_1, device=cpu)
|
||||
# lif learn
|
||||
lifLearn!(kfn.lif_wRec,
|
||||
kfn.lif_wRecChange,
|
||||
kfn.lif_arrayProjection4d)
|
||||
kfn.lif_arrayProjection4d,
|
||||
kfn.lif_neuronInactivityCounter,
|
||||
kfn.lif_synapticConnectionNumber,
|
||||
kfn.zit_cumulative,
|
||||
device)
|
||||
|
||||
# alif learn
|
||||
alifLearn!(kfn.alif_wRec,
|
||||
kfn.alif_wRecChange,
|
||||
kfn.alif_arrayProjection4d)
|
||||
kfn.alif_arrayProjection4d,
|
||||
kfn.alif_neuronInactivityCounter,
|
||||
kfn.alif_synapticConnectionNumber,
|
||||
kfn.zit_cumulative,
|
||||
device)
|
||||
|
||||
# on learn
|
||||
onLearn!(kfn.on_wOut,
|
||||
@@ -289,23 +297,52 @@ end
|
||||
|
||||
function lifLearn!(wRec,
|
||||
wRecChange,
|
||||
arrayProjection4d)
|
||||
# merge learning weight with average learning weight
|
||||
arrayProjection4d,
|
||||
inactivityCounter,
|
||||
synapticConnectionNumber,
|
||||
zit_cumulative,
|
||||
device)
|
||||
# merge learning weight with average learning weight of all batch
|
||||
wRec .+= (sum(wRecChange, dims=4) ./ (size(wRec, 4))) .* arrayProjection4d
|
||||
|
||||
#TODO synaptic strength
|
||||
wRec_cpu = wRec |> cpu
|
||||
wRec_cpu = wRec_cpu[:,:,:,1] # since every batch has the same neuron wRec, (row, col, n)
|
||||
inactivityCounter_cpu = inactivityCounter |> cpu
|
||||
inactivityCounter_cpu = inactivityCounter_cpu[:,:,:,1] # (row, col, n)
|
||||
zit_cumulative_cpu = zit_cumulative |> cpu
|
||||
zit_cumulative_cpu = zit_cumulative_cpu[:,:,1] # (row, col)
|
||||
|
||||
#TODO neuroplasticity
|
||||
# weak / negative synaptic connection will get randomed in neuroplasticity()
|
||||
wRec_cpu = GeneralUtils.replaceBetween.(wRec_cpu, 0.0, 0.1, -1.0) # mark with -1.0
|
||||
|
||||
# synaptic connection that has no inactivity will get randomed in neuroplasticity()
|
||||
GeneralUtils.replace_elements!(inactivityCounter_cpu, 0.0, wRec_cpu, -1.0)
|
||||
# reset lif_inactivity elements to 10000
|
||||
GeneralUtils.replace_elements!(inactivityCounter_cpu, 0.0, -9.0) # -9.0 is base value
|
||||
|
||||
#WORKING neuroplasticity
|
||||
wRec_cpu = neuroplasticity(synapticConnectionNumber, zit_cumulative_cpu, wRec_cpu,
|
||||
inactivityCounter_cpu)
|
||||
error("DEBUG -> lifLearn! $(Dates.now())")
|
||||
# #TODO send to device with correct dimension
|
||||
# wRec = wRec |> device
|
||||
# inactivityCounter = inactivityCounter_cpu |> device
|
||||
|
||||
# error("DEBUG -> lifLearn! $(Dates.now())")
|
||||
end
|
||||
|
||||
function alifLearn!(wRec,
|
||||
wRecChange,
|
||||
arrayProjection4d)
|
||||
arrayProjection4d,
|
||||
inactivityCounter,
|
||||
synapticConnectionNumber,
|
||||
zit_cumulative,
|
||||
device)
|
||||
# merge learning weight with average learning weight
|
||||
wRec .+= (sum(wRecChange, dims=4) ./ (size(wRec, 4))) .* arrayProjection4d
|
||||
|
||||
# weak / negative synaptic connection will get randomed in neuroplasticity()
|
||||
wRec .= GeneralUtils.replaceLessThan.(wRec, 0.01, 0.0)
|
||||
|
||||
#TODO synaptic strength
|
||||
|
||||
#TODO neuroplasticity
|
||||
@@ -327,8 +364,56 @@ function onLearn!(wOut,
|
||||
|
||||
end
|
||||
|
||||
function neuroplasticity(synapticConnectionNumber,
|
||||
zit_cumulative, # (row, col)
|
||||
wRec, # (row, col, n)
|
||||
inactivityCounter_cpu) # (row, col, n)
|
||||
|
||||
i1,i2,i3 = size(wRec)
|
||||
|
||||
# for each neuron, find total number of synaptic conn that should draw
|
||||
# new connection to firing and non-firing neurons pool
|
||||
subToFireNeuron_toBe = Int(floor(0.7 * synapticConnectionNumber))
|
||||
subToNonFiringNeuron_toBe = synapticConnectionNumber - subToFireNeuron_toBe
|
||||
|
||||
#WORKING for each neuron, count how many synap already subscribed to firing-neurons
|
||||
subToFireNeuron_current = sum((!iszero).(zit_cumulative .* wRec), dims=(1,2)) # (1, 1, n)
|
||||
subToNonFiringNeuron_current = synapticConnectionNumber .- subToFireNeuron_current # (1, 1, n)
|
||||
mask = (!iszero).(zit_cumulative) # mask of firing neurons = 1, non-firing = 0
|
||||
projection = ones(i1,i2,i3)
|
||||
mask = mask .* projection # (row, col, n)
|
||||
totalNewConn = sum(isequal.(wRec, -1.0), dims=(1,2)) # count new conn mark (-1.0), (1, 1, n)
|
||||
println("mask ", size(mask))
|
||||
println("wRec ", size(wRec))
|
||||
println("inactivityCounter_cpu ", size(inactivityCounter_cpu))
|
||||
println("totalNeurons ", totalNewConn, size(totalNewConn))
|
||||
error("DEBUG -> neuroplasticity $(Dates.now())")
|
||||
for i in 1:i3
|
||||
|
||||
|
||||
# add new conn to firing neurons pool
|
||||
remaining = GeneralUtils.replace_elements(mask[:,:,i],
|
||||
1,
|
||||
wRecmask[:,:,i],
|
||||
inactivityCounter_cpumask[:,:,i],
|
||||
totalNewConn[:,:,i])
|
||||
|
||||
#TODO add new conn to non-firing neurons pool
|
||||
|
||||
end
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
newFiringConn = subToFireNeuron_toBe - subToFireNeuron_current
|
||||
newFiringConn = newFiringConn > 0 ? newFiringConn : 0
|
||||
|
||||
newNonFiringConn = subToNonFiringNeuron_toBe - subToNonFiringNeuron_current
|
||||
|
||||
return wRec
|
||||
end
|
||||
|
||||
|
||||
|
||||
|
||||
@@ -21,6 +21,23 @@ function refractoryStatus!(refractoryCounter, refractoryActive, refractoryInacti
|
||||
end
|
||||
end
|
||||
|
||||
function frobenius_distance(A, B)
|
||||
# Check if the matrices have the same size
|
||||
if size(A) != size(B)
|
||||
error("The matrices must have the same size")
|
||||
end
|
||||
# Initialize the distance to zero
|
||||
distance = 0.0
|
||||
# Loop over the elements of the matrices and add the squared differences
|
||||
for i in 1:size(A, 1)
|
||||
for j in 1:size(A, 2)
|
||||
distance += (A[i, j] - B[i, j])^2
|
||||
end
|
||||
end
|
||||
# Return the square root of the distance
|
||||
return sqrt(distance)
|
||||
end
|
||||
|
||||
|
||||
|
||||
|
||||
@@ -59,13 +76,6 @@ end
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
240
src/type.jl
240
src/type.jl
@@ -24,6 +24,7 @@ Base.@kwdef mutable struct kfn_1 <: knowledgeFn
|
||||
inputSize::Union{AbstractArray, Nothing} = nothing
|
||||
zit::Union{AbstractArray, Nothing} = nothing # 3D activation matrix
|
||||
zit_cumulative::Union{AbstractArray, Nothing} = nothing
|
||||
exInType::Union{AbstractArray, Nothing} = nothing
|
||||
modelError::Union{AbstractArray, Nothing} = nothing # store RSNN error
|
||||
outputError::Union{AbstractArray, Nothing} = nothing # store output neurons error
|
||||
|
||||
@@ -52,14 +53,17 @@ Base.@kwdef mutable struct kfn_1 <: knowledgeFn
|
||||
lif_gammaPd::Union{AbstractArray, Nothing} = nothing
|
||||
lif_wRecChange::Union{AbstractArray, Nothing} = nothing
|
||||
lif_error::Union{AbstractArray, Nothing} = nothing
|
||||
lif_subscription::Union{AbstractArray, Nothing} = nothing
|
||||
|
||||
lif_firingCounter::Union{AbstractArray, Nothing} = nothing
|
||||
lif_firingTargetFrequency::Union{AbstractArray, Nothing} = nothing
|
||||
lif_neuronInactivityCounter::Union{AbstractArray, Nothing} = nothing
|
||||
lif_synapticInactivityCounter::Union{AbstractArray, Nothing} = nothing
|
||||
lif_synapticConnectionNumber::Union{Int, Nothing} = nothing
|
||||
|
||||
# pre-allocation array
|
||||
lif_arrayProjection4d::Union{AbstractArray, Nothing} = nothing # use to project 3d array to 4d
|
||||
lif_recSignal::Union{AbstractArray, Nothing} = nothing
|
||||
lif_exInType::Union{AbstractArray, Nothing} = nothing
|
||||
# lif_decayed_epsilonRec::Union{AbstractArray, Nothing} = nothing
|
||||
# lif_vt_diff_vth::Union{AbstractArray, Nothing} = nothing
|
||||
# lif_vt_diff_vth_div_vth::Union{AbstractArray, Nothing} = nothing
|
||||
@@ -89,14 +93,17 @@ Base.@kwdef mutable struct kfn_1 <: knowledgeFn
|
||||
alif_gammaPd::Union{AbstractArray, Nothing} = nothing
|
||||
alif_wRecChange::Union{AbstractArray, Nothing} = nothing
|
||||
alif_error::Union{AbstractArray, Nothing} = nothing
|
||||
alif_subscription::Union{AbstractArray, Nothing} = nothing
|
||||
|
||||
alif_firingCounter::Union{AbstractArray, Nothing} = nothing
|
||||
alif_firingTargetFrequency::Union{AbstractArray, Nothing} = nothing
|
||||
alif_neuronInactivityCounter::Union{AbstractArray, Nothing} = nothing
|
||||
alif_synapticInactivityCounter::Union{AbstractArray, Nothing} = nothing
|
||||
alif_synapticConnectionNumber::Union{Int, Nothing} = nothing
|
||||
|
||||
# pre-allocation array
|
||||
alif_arrayProjection4d::Union{AbstractArray, Nothing} = nothing # use to project 3d array to 4d
|
||||
alif_recSignal::Union{AbstractArray, Nothing} = nothing
|
||||
alif_exInType::Union{AbstractArray, Nothing} = nothing
|
||||
# alif_decayed_epsilonRec::Union{AbstractArray, Nothing} = nothing
|
||||
# alif_vt_diff_vth::Union{AbstractArray, Nothing} = nothing
|
||||
# alif_vt_diff_vth_div_vth::Union{AbstractArray, Nothing} = nothing
|
||||
@@ -168,15 +175,17 @@ function kfn_1(params::Dict; device=cpu)
|
||||
# initialize activation matrix #
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# row*col is a 2D matrix represent all RSNN activation
|
||||
row, col, batch = kfn.params[:inputPort][:signal][:numbers] # z-axis represent signal batch number
|
||||
row, signal_col, batch = kfn.params[:inputPort][:signal][:numbers] # z-axis represent signal batch number
|
||||
|
||||
kfn.inputSize = [row, col] |> device
|
||||
col += kfn.params[:computeNeuron][:lif][:numbers][2]
|
||||
col += kfn.params[:computeNeuron][:alif][:numbers][2]
|
||||
kfn.inputSize = [row, signal_col] |> device
|
||||
lif_col = kfn.params[:computeNeuron][:lif][:numbers][2]
|
||||
alif_col = kfn.params[:computeNeuron][:alif][:numbers][2]
|
||||
|
||||
col = signal_col + lif_col + alif_col
|
||||
|
||||
# activation matrix
|
||||
kfn.zit = zeros(row, col, batch) |> device
|
||||
kfn.zit_cumulative = (similar(kfn.zit) .= 0) |> device
|
||||
kfn.zit_cumulative = (similar(kfn.zit) .= 0)
|
||||
kfn.modelError = zeros(1) |> device
|
||||
|
||||
# ---------------------------------------------------------------------------- #
|
||||
@@ -184,115 +193,133 @@ function kfn_1(params::Dict; device=cpu)
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# In 3D LIF matrix, z-axis represent each neuron while each 2D slice represent that neuron's
|
||||
# synaptic subscription to other neurons (via activation matrix)
|
||||
n = kfn.params[:computeNeuron][:lif][:numbers][1] * kfn.params[:computeNeuron][:lif][:numbers][2]
|
||||
lif_n = kfn.params[:computeNeuron][:lif][:numbers][1] * kfn.params[:computeNeuron][:lif][:numbers][2]
|
||||
|
||||
# subscription
|
||||
w = zeros(row, col, n)
|
||||
w = zeros(row, col, lif_n)
|
||||
synapticConnectionPercent = kfn.params[:computeNeuron][:lif][:params][:synapticConnectionPercent]
|
||||
synapticConnection = Int(floor(row*col * synapticConnectionPercent/100))
|
||||
kfn.lif_synapticConnectionNumber = Int(floor(row*col * synapticConnectionPercent/100))
|
||||
for slice in eachslice(w, dims=3)
|
||||
pool = shuffle!([1:row*col...])[1:synapticConnection]
|
||||
pool = shuffle!([1:row*col...])[1:kfn.lif_synapticConnectionNumber]
|
||||
for i in pool
|
||||
slice[i] = randn()/10 # assign weight to synaptic connection. /10 to start small,
|
||||
slice[i] = rand() # assign weight to synaptic connection. /10 to start small,
|
||||
# otherwise RSNN's vt Usually stay negative (-)
|
||||
end
|
||||
end
|
||||
|
||||
# 10% of neuron connection should be enough to start to make neuron fires
|
||||
should_be_avg_weight = 1 / (0.1 * lif_n)
|
||||
w = w .* (should_be_avg_weight / maximum(w)) # adjust overall weight
|
||||
|
||||
# project 3D w into 4D kfn.lif_wRec (row, col, n, batch)
|
||||
kfn.lif_wRec = reshape(w, (row, col, n, 1)) .* ones(row, col, n, batch) |> device
|
||||
kfn.lif_zit = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
kfn.lif_vt = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
kfn.lif_vth = (similar(kfn.lif_wRec) .= 1) |> device
|
||||
kfn.lif_vRest = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
kfn.lif_zt = zeros(1, 1, n, batch) |> device
|
||||
kfn.lif_zt4d = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
kfn.lif_refractoryCounter = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
kfn.lif_refractoryDuration = (similar(kfn.lif_wRec) .= 3) |> device
|
||||
kfn.lif_wRec = reshape(w, (row, col, lif_n, 1)) .* ones(row, col, lif_n, batch) |> device
|
||||
kfn.lif_zit = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_vt = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_vth = (similar(kfn.lif_wRec) .= 1)
|
||||
kfn.lif_vRest = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_zt = zeros(1, 1, lif_n, batch) |> device
|
||||
kfn.lif_zt4d = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_refractoryCounter = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_refractoryDuration = (similar(kfn.lif_wRec) .= 3)
|
||||
kfn.lif_delta = 1.0
|
||||
kfn.lif_tau_m = 20.0
|
||||
kfn.lif_alpha = (similar(kfn.lif_wRec) .= (exp(-kfn.lif_delta / kfn.lif_tau_m))) |> device
|
||||
kfn.lif_phi = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
kfn.lif_epsilonRec = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
kfn.lif_eRec = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
kfn.lif_eta = (similar(kfn.lif_wRec) .= 0.001) |> device
|
||||
kfn.lif_gammaPd = (similar(kfn.lif_wRec) .= 0.3) |> device
|
||||
kfn.lif_wRecChange = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
kfn.lif_error = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
kfn.lif_subscription = (GeneralUtils.isNotEqual.(kfn.lif_wRec, 0)) |> device
|
||||
kfn.lif_alpha = (similar(kfn.lif_wRec) .= (exp(-kfn.lif_delta / kfn.lif_tau_m)))
|
||||
kfn.lif_phi = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_epsilonRec = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_eRec = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_eta = (similar(kfn.lif_wRec) .= 0.001)
|
||||
kfn.lif_gammaPd = (similar(kfn.lif_wRec) .= 0.3)
|
||||
kfn.lif_wRecChange = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_error = (similar(kfn.lif_wRec) .= 0)
|
||||
|
||||
kfn.lif_firingCounter = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
# firingTargetFrequency = desired count / total sequence length
|
||||
kfn.lif_firingTargetFrequency = (similar(kfn.lif_wRec) .= 0.1) |> device
|
||||
kfn.lif_firingCounter = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_firingTargetFrequency = (similar(kfn.lif_wRec) .= 0.1)
|
||||
kfn.lif_neuronInactivityCounter = (similar(kfn.lif_wRec) .= 10000)
|
||||
kfn.lif_synapticInactivityCounter = Array(similar(kfn.lif_wRec) .= -9) # -9 for non-sub conn
|
||||
mask = Array((!iszero).(kfn.lif_wRec))
|
||||
GeneralUtils.replace_elements!(mask, 1, kfn.lif_synapticInactivityCounter, 10000)
|
||||
kfn.lif_synapticInactivityCounter = kfn.lif_synapticInactivityCounter |> device
|
||||
|
||||
kfn.lif_arrayProjection4d = (similar(kfn.lif_wRec) .= 1) |> device
|
||||
kfn.lif_recSignal = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
# kfn.lif_decayed_epsilonRec = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
# kfn.lif_vt_diff_vth = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
# kfn.lif_vt_diff_vth_div_vth = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
# kfn.lif_gammaPd_div_vth = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
# kfn.lif_phiActivation = (similar(kfn.lif_wRec) .= 0) |> device
|
||||
kfn.lif_arrayProjection4d = (similar(kfn.lif_wRec) .= 1)
|
||||
kfn.lif_recSignal = (similar(kfn.lif_wRec) .= 0)
|
||||
kfn.lif_exInType = (similar(kfn.lif_wRec) .= 0)
|
||||
# kfn.lif_decayed_epsilonRec = (similar(kfn.lif_wRec) .= 0)
|
||||
# kfn.lif_vt_diff_vth = (similar(kfn.lif_wRec) .= 0)
|
||||
# kfn.lif_vt_diff_vth_div_vth = (similar(kfn.lif_wRec) .= 0)
|
||||
# kfn.lif_gammaPd_div_vth = (similar(kfn.lif_wRec) .= 0)
|
||||
# kfn.lif_phiActivation = (similar(kfn.lif_wRec) .= 0)
|
||||
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# ALIF config #
|
||||
# ---------------------------------------------------------------------------- #
|
||||
n = kfn.params[:computeNeuron][:alif][:numbers][1] * kfn.params[:computeNeuron][:alif][:numbers][2]
|
||||
alif_n = kfn.params[:computeNeuron][:alif][:numbers][1] * kfn.params[:computeNeuron][:alif][:numbers][2]
|
||||
|
||||
# subscription
|
||||
w = zeros(row, col, n)
|
||||
w = zeros(row, col, alif_n)
|
||||
synapticConnectionPercent = kfn.params[:computeNeuron][:alif][:params][:synapticConnectionPercent]
|
||||
synapticConnection = Int(floor(row*col * synapticConnectionPercent/100))
|
||||
kfn.alif_synapticConnectionNumber = Int(floor(row*col * synapticConnectionPercent/100))
|
||||
for slice in eachslice(w, dims=3)
|
||||
pool = shuffle!([1:row*col...])[1:synapticConnection]
|
||||
pool = shuffle!([1:row*col...])[1:kfn.alif_synapticConnectionNumber]
|
||||
for i in pool
|
||||
slice[i] = randn()/10 # assign weight to synaptic connection. /10 to start small,
|
||||
slice[i] = rand() # assign weight to synaptic connection. /10 to start small,
|
||||
# otherwise RSNN's vt Usually stay negative (-)
|
||||
end
|
||||
end
|
||||
|
||||
# 10% of neuron connection should be enough to start to make neuron fires
|
||||
should_be_avg_weight = 1 / (0.1 * alif_n)
|
||||
w = w .* (should_be_avg_weight / maximum(w)) # adjust overall weight
|
||||
|
||||
# project 3D w into 4D kfn.alif_wRec
|
||||
kfn.alif_wRec = reshape(w, (row, col, n, 1)) .* ones(row, col, n, batch) |> device
|
||||
kfn.alif_zit = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
kfn.alif_vt = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
kfn.alif_vth = (similar(kfn.alif_wRec) .= 1) |> device
|
||||
kfn.alif_vRest = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
kfn.alif_zt = zeros(1, 1, n, batch) |> device
|
||||
kfn.alif_zt4d = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
kfn.alif_refractoryCounter = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
kfn.alif_refractoryDuration = (similar(kfn.alif_wRec) .= 3) |> device
|
||||
kfn.alif_wRec = reshape(w, (row, col, alif_n, 1)) .* ones(row, col, alif_n, batch) |> device
|
||||
kfn.alif_zit = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_vt = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_vth = (similar(kfn.alif_wRec) .= 1)
|
||||
kfn.alif_vRest = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_zt = zeros(1, 1, alif_n, batch) |> device
|
||||
kfn.alif_zt4d = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_refractoryCounter = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_refractoryDuration = (similar(kfn.alif_wRec) .= 3)
|
||||
kfn.alif_delta = 1.0
|
||||
kfn.alif_tau_m = 20.0
|
||||
kfn.alif_alpha = (similar(kfn.alif_wRec) .= (exp(-kfn.alif_delta / kfn.alif_tau_m))) |> device
|
||||
kfn.alif_phi = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
kfn.alif_epsilonRec = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
kfn.alif_eRec = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
kfn.alif_eta = (similar(kfn.alif_wRec) .= 0.001) |> device
|
||||
kfn.alif_gammaPd = (similar(kfn.alif_wRec) .= 0.3) |> device
|
||||
kfn.alif_wRecChange = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
kfn.alif_error = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
kfn.alif_subscription = (GeneralUtils.isNotEqual.(kfn.alif_wRec, 0)) |> device
|
||||
kfn.alif_alpha = (similar(kfn.alif_wRec) .= (exp(-kfn.alif_delta / kfn.alif_tau_m)))
|
||||
kfn.alif_phi = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_epsilonRec = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_eRec = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_eta = (similar(kfn.alif_wRec) .= 0.001)
|
||||
kfn.alif_gammaPd = (similar(kfn.alif_wRec) .= 0.3)
|
||||
kfn.alif_wRecChange = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_error = (similar(kfn.alif_wRec) .= 0)
|
||||
|
||||
kfn.alif_firingCounter = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
# firingTargetFrequency = desired count / total sequence length
|
||||
kfn.alif_firingTargetFrequency = (similar(kfn.alif_wRec) .= 0.1) |> device
|
||||
kfn.alif_firingCounter = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_firingTargetFrequency = (similar(kfn.alif_wRec) .= 0.1)
|
||||
kfn.alif_neuronInactivityCounter = (similar(kfn.alif_wRec) .= 10000)
|
||||
kfn.alif_synapticInactivityCounter = Array(similar(kfn.alif_wRec) .= -9) # -9 for non-sub conn
|
||||
mask = Array((!iszero).(kfn.alif_wRec))
|
||||
GeneralUtils.replace_elements!(mask, 1, kfn.alif_synapticInactivityCounter, 10000)
|
||||
kfn.alif_synapticInactivityCounter = kfn.alif_synapticInactivityCounter |> device
|
||||
|
||||
kfn.alif_arrayProjection4d = (similar(kfn.alif_wRec) .= 1) |> device
|
||||
kfn.alif_recSignal = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
# kfn.alif_decayed_epsilonRec = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
# kfn.alif_vt_diff_vth = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
# kfn.alif_vt_diff_vth_div_vth = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
# kfn.alif_gammaPd_div_vth = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
# kfn.alif_phiActivation = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
kfn.alif_arrayProjection4d = (similar(kfn.alif_wRec) .= 1)
|
||||
kfn.alif_recSignal = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_exInType = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_decayed_epsilonRec = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_vt_diff_vth = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_vt_diff_vth_div_vth = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_gammaPd_div_vth = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_phiActivation = (similar(kfn.alif_wRec) .= 0)
|
||||
|
||||
# alif specific variables
|
||||
kfn.alif_epsilonRecA = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
kfn.alif_avth = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
kfn.alif_a = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
kfn.alif_beta = (similar(kfn.alif_wRec) .= 0.07) |> device
|
||||
kfn.alif_epsilonRecA = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_avth = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_a = (similar(kfn.alif_wRec) .= 0)
|
||||
kfn.alif_beta = (similar(kfn.alif_wRec) .= 0.07)
|
||||
kfn.alif_tau_a = 800.0
|
||||
kfn.alif_rho = (similar(kfn.alif_wRec) .= (exp(-kfn.alif_delta / kfn.alif_tau_a))) |> device
|
||||
# kfn.alif_phi_x_epsilonRec = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
# kfn.alif_phi_x_beta = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
# kfn.alif_rho_diff_phi_x_beta = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
# kfn.alif_rho_div_phi_x_beta_x_epsilonRecA = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
# kfn.alif_beta_x_a = (similar(kfn.alif_wRec) .= 0) |> device
|
||||
# kfn.alif_phi_x_epsilonRec = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_phi_x_beta = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_rho_diff_phi_x_beta = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_rho_div_phi_x_beta_x_epsilonRecA = (similar(kfn.alif_wRec) .= 0)
|
||||
# kfn.alif_beta_x_a = (similar(kfn.alif_wRec) .= 0)
|
||||
|
||||
# ---------------------------------------------------------------------------- #
|
||||
# output config #
|
||||
@@ -310,38 +337,49 @@ function kfn_1(params::Dict; device=cpu)
|
||||
# pool must contain only lif, alif neurons
|
||||
pool = shuffle!([startInd:row*col...])[1:synapticConnection]
|
||||
for i in pool
|
||||
slice[i] = randn()/10 # assign weight to synaptic connection. /10 to start small,
|
||||
slice[i] = rand() # assign weight to synaptic connection. /10 to start small,
|
||||
# otherwise RSNN's vt Usually stay negative (-)
|
||||
end
|
||||
end
|
||||
|
||||
# # 10% of neuron connection should be enough to start to make neuron fires
|
||||
# should_be_avg_weight = 1 / (0.2 * n)
|
||||
# w = w .* (should_be_avg_weight / maximum(w)) # adjust overall weight
|
||||
|
||||
# project 3D w into 4D kfn.lif_wOut (row, col, n, batch)
|
||||
kfn.on_wOut = reshape(w, (row, col, n, 1)) .* ones(row, col, n, batch) |> device
|
||||
kfn.on_zit = (similar(kfn.on_wOut) .= 0) |> device
|
||||
kfn.on_vt = (similar(kfn.on_wOut) .= 0) |> device
|
||||
kfn.on_vth = (similar(kfn.on_wOut) .= 1) |> device
|
||||
kfn.on_vRest = (similar(kfn.on_wOut) .= 0) |> device
|
||||
kfn.on_zit = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_vt = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_vth = (similar(kfn.on_wOut) .= 1)
|
||||
kfn.on_vRest = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_zt = zeros(1, 1, n, batch) |> device
|
||||
kfn.on_zt4d = (similar(kfn.on_wOut) .= 0) |> device
|
||||
kfn.on_refractoryCounter = (similar(kfn.on_wOut) .= 0) |> device
|
||||
kfn.on_refractoryDuration = (similar(kfn.on_wOut) .= 0) |> device
|
||||
kfn.on_zt4d = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_refractoryCounter = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_refractoryDuration = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_delta = 1.0
|
||||
kfn.on_tau_m = 20.0
|
||||
kfn.on_alpha = (similar(kfn.on_wOut) .= (exp(-kfn.on_delta / kfn.on_tau_m))) |> device
|
||||
kfn.on_phi = (similar(kfn.on_wOut) .= 0) |> device
|
||||
kfn.on_epsilonRec = (similar(kfn.on_wOut) .= 0) |> device
|
||||
kfn.on_eRec = (similar(kfn.on_wOut) .= 0) |> device
|
||||
kfn.on_eta = (similar(kfn.on_wOut) .= 0.001) |> device
|
||||
kfn.on_gammaPd = (similar(kfn.on_wOut) .= 0.3) |> device
|
||||
kfn.on_wOutChange = (similar(kfn.on_wOut) .= 0) |> device
|
||||
kfn.on_error = (similar(kfn.on_wOut) .= 0) |> device
|
||||
kfn.on_alpha = (similar(kfn.on_wOut) .= (exp(-kfn.on_delta / kfn.on_tau_m)))
|
||||
kfn.on_phi = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_epsilonRec = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_eRec = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_eta = (similar(kfn.on_wOut) .= 0.001)
|
||||
kfn.on_gammaPd = (similar(kfn.on_wOut) .= 0.3)
|
||||
kfn.on_wOutChange = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_error = (similar(kfn.on_wOut) .= 0)
|
||||
kfn.on_subscription = (GeneralUtils.isNotEqual.(kfn.on_wOut, 0)) |> device
|
||||
|
||||
kfn.on_firingCounter = (similar(kfn.on_wOut) .= 0) |> device
|
||||
kfn.on_firingCounter = (similar(kfn.on_wOut) .= 0)
|
||||
|
||||
kfn.on_arrayProjection4d = (similar(kfn.on_wOut) .= 1) |> device
|
||||
kfn.on_recSignal = (similar(kfn.on_wOut) .= 0) |> device
|
||||
kfn.on_arrayProjection4d = (similar(kfn.on_wOut) .= 1)
|
||||
kfn.on_recSignal = (similar(kfn.on_wOut) .= 0)
|
||||
|
||||
kfn.outputError = zeros(n, batch) |> device
|
||||
totalComputeNeurons = lif_n + alif_n
|
||||
inhabitoryNeurons = Int(floor(totalComputeNeurons * 30/100))
|
||||
mask1 = ones(row, signal_col)
|
||||
mask2 = GeneralUtils.multiply_random_elements(ones(row, lif_col + alif_col),
|
||||
-1, inhabitoryNeurons, MersenneTwister(1234))
|
||||
kfn.exInType = cat(mask1, mask2, dims=2) |> device
|
||||
|
||||
return kfn
|
||||
end
|
||||
|
||||
Reference in New Issue
Block a user