Module: Neuronet::Squash

Included in:

InputNeuron, MiddleNeuron, Neuron

Defined in:

lib/neuronet/squash.rb

Overview

Squash provides logistic sigmoid function.

Class Method Summary

• .squash(value) ⇒ Object

  Logistic sigmoid: maps Real to (0, 1).

• .unsquash(activation) ⇒ Object

  Inverse sigmoid: maps (0, 1) to Real.

Class Method Details

.squash(value) ⇒ Object

Logistic sigmoid: maps Real to (0, 1).

# File 'lib/neuronet/squash.rb', line 7

def squash(value) = 1.0 / (1.0 + Math.exp(-value))
# Inverse sigmoid: maps (0, 1) to Real.

.unsquash(activation) ⇒ Object

Inverse sigmoid: maps (0, 1) to Real.

# File 'lib/neuronet/squash.rb', line 9

def unsquash(activation) = Math.log(activation / (1.0 - activation))