Skip to contents

Scaled ELU with the Klambauer self-normalising constants. Note the branch is z <= 0 here (not z < 0), matching the Python module.

Usage

morie_geron_selu(z, lam = 1.05070098735548, alpha = 1.67326324235438)

Arguments

z

Numeric vector.

lam, alpha

Positive scale constants.

Value

List with a, grad, mean, var (population), lam, alpha, estimate, n.

Examples

V <- c(1, 2, 3, 4, 5, 6, 7, 8)
morie_geron_selu(V)
#> $a
#> [1] 1.050701 2.101402 3.152103 4.202804 5.253505 6.304206 7.354907 8.405608
#> 
#> $grad
#> [1] 1.050701 1.050701 1.050701 1.050701 1.050701 1.050701 1.050701 1.050701
#> 
#> $mean
#> [1] 4.728154
#> 
#> $var
#> [1] 5.795856
#> 
#> $lam
#> [1] 1.050701
#> 
#> $alpha
#> [1] 1.673263
#> 
#> $estimate
#> [1] 4.728154
#> 
#> $n
#> [1] 8
#> 
#> $method
#> [1] "Scaled ELU with the Klambauer self-normalising constants"
#>