TastyPancakes/Flux.jl
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Loss function names in lowercase

Browse Source
This commit is contained in:
thebhatman 2019-03-26 03:09:48 +05:30
parent 57a52e3375
commit c4d12e57fe
2 changed files with 12 additions and 15 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
src/layers/stateless.jl
View File

@ -59,9 +59,8 @@ end
Kullback Leibler Divergence(KL Divergence) Kullback Leibler Divergence(KL Divergence)
KLDivergence is a measure of how much one probability distribution is different from the other. KLDivergence is a measure of how much one probability distribution is different from the other.
It is always non-negative and zero only when both the distributions are equal everywhere. It is always non-negative and zero only when both the distributions are equal everywhere.
""" """
function KLDivergence(ŷ, y) function kldivergence(ŷ, y)
entropy = sum(y .* log.(y)) *1 //size(y,2) entropy = sum(y .* log.(y)) *1 //size(y,2)
cross_entropy = crossentropy(ŷ, y) cross_entropy = crossentropy(ŷ, y)
return entropy + cross_entropy return entropy + cross_entropy
@ -70,15 +69,13 @@ end
""" """
Poisson Loss function Poisson Loss function
Poisson loss function is a measure of how the predicted distribution diverges from the expected distribution. Poisson loss function is a measure of how the predicted distribution diverges from the expected distribution.
""" """
Poisson(ŷ, y) = sum(ŷ .- y .* log.(ŷ)) *1 // size(y,2) poisson(ŷ, y) = sum(ŷ .- y .* log.(ŷ)) *1 // size(y,2)
""" """
Logcosh Loss function Logcosh Loss function
""" """
logcosh(ŷ, y) = sum(log.(cosh.(ŷ .- y))) logcosh(ŷ, y) = sum(log.(cosh.(ŷ .- y)))
Hinge(ŷ, y) = sum(max.(0.0, 1 .- ŷ .* y)) *1 // size(y,2) hinge(ŷ, y) = sum(max.(0.0, 1 .- ŷ .* y)) *1 // size(y,2)

18
test/layers/stateless.jl
View File

@ -52,23 +52,23 @@ const ϵ = 1e-7
y = [1 2 3] y = [1 2 3]
y1 = [4.0 5.0 6.0] y1 = [4.0 5.0 6.0]
@testset "KLDivergence" begin @testset "kldivergence" begin
@test Flux.KLDivergence(y, y1) ≈ 4.761838062403337 @test Flux.kldivergence(y, y1) ≈ 4.761838062403337
@test Flux.KLDivergence(y, y) ≈ 0 @test Flux.kldivergence(y, y) ≈ 0
end end
y = [1 2 3 4] y = [1 2 3 4]
y1 = [5.0 6.0 7.0 8.0] y1 = [5.0 6.0 7.0 8.0]
@testset "Hinge" begin @testset "hinge" begin
@test Flux.Hinge(y, y1) ≈ 0 @test Flux.hinge(y, y1) ≈ 0
@test Flux.Hinge(y, 0.5 .* y) ≈ 0.125 @test Flux.hinge(y, 0.5 .* y) ≈ 0.125
end end
y = [0.1 0.2 0.3] y = [0.1 0.2 0.3]
y1 = [0.4 0.5 0.6] y1 = [0.4 0.5 0.6]
@testset "Poisson" begin @testset "poisson" begin
@test Flux.Poisson(y, y1) ≈ 1.0160455586700767 @test Flux.poisson(y, y1) ≈ 1.0160455586700767
@test Flux.Poisson(y, y) ≈ 0.5044459776946685 @test Flux.poisson(y, y) ≈ 0.5044459776946685
end end
@testset "logcosh" begin @testset "logcosh" begin