forked from rosh/NeuroBook
40 lines
1.1 KiB
Python
40 lines
1.1 KiB
Python
 | a=1.0
| ||
b=0.0
| |||
theta=0
| |||
# Step activation function
| |||
# Constant 'theta' determines the level of neuron activation.
| |||
# Parameter 'x' Weighted sum of initial data.
| |||
def ActStep (x):
| |||
return 1 if x>=theta else 0
| |||
| |||
# Linear activation function
| |||
# Constant 'a' defines the angle of inclination of the line, and 'b' - the vertical offset of the line
| |||
# Parameter 'x' Weighted sum of initial data.
| |||
def ActLinear (x):
| |||
return a*x+b
| |||
| |||
# Sigmoid activation function
| |||
# Constant 'a' stretches the range of values of the function from '0' to 'a'
| |||
# Constant 'b' shifts the resulting value
| |||
# Parameter 'x' Weighted sum of initial data.
| |||
import math
| |||
def ActSigmoid(x):
| |||
return a/(1+math.exp(-x))-b
| |||
| |||
# TANH activation function
| |||
# Parameter 'x' Weighted sum of initial data.
| |||
import math
| |||
def ActTanh (x):
| |||
return math.tanh(x)
| |||
| |||
# PReLU activation function
| |||
# Constant 'a' leak parameter
| |||
# Parameter 'x' Weighted sum of initial data.
| |||
def ActPReLU (x):
| |||
return x if x>=0 else a*x
| |||
| |||
# SoftMax activation function
| |||
# Parameter 'X' array of weighted initial data.
| |||
from scipy.special import softmax
| |||
def ActSoftMax (X):
| |||
return softmax(X)
|