Author - Mohit Rathore mrmohitrathoremr@gmail.com - markroxor.in
Licensed under The MIT License - https://opensource.org/licenses/MIT

In [1]:
from fromscratchtoml.neural_network import Activations
from fromscratchtoml.toolbox import binary_visualize

import numpy as np

# set this to avoid exponential notations
np.set_printoptions(suppress=True, formatter={'float': '{: 0.8f}'.format})
%matplotlib inline

Visualising activation funtions

Linear Activation

In [3]:
X = np.linspace(-5,5,20)
Y = Activations.linear(X)
XY = np.stack([X, Y], axis=1)
print(repr(Y))
binary_visualize(XY)

array([-5.00000000, -4.47368421, -3.94736842, -3.42105263, -2.89473684,
       -2.36842105, -1.84210526, -1.31578947, -0.78947368, -0.26315789,
        0.26315789,  0.78947368,  1.31578947,  1.84210526,  2.36842105,
        2.89473684,  3.42105263,  3.94736842,  4.47368421,  5.00000000])

tanh Activation

In [4]:
X = np.linspace(-5,5,20)
Y = Activations.tanh(X)
XY = np.stack([X, Y], axis=1)
print(repr(Y))
binary_visualize(XY)

array([-0.9999092 , -0.99973988, -0.99925488, -0.99786657, -0.99389948,
       -0.98261979, -0.95099682, -0.865733  , -0.65811078, -0.25724684,
        0.25724684,  0.65811078,  0.865733  ,  0.95099682,  0.98261979,
        0.99389948,  0.99786657,  0.99925488,  0.99973988,  0.9999092 ],
      dtype=float128)

sigmoid activation

In [5]:
X = np.linspace(-5,5,20)
Y = Activations.sigmoid(X)
XY = np.stack([X, Y], axis=1)
print(repr(Y))
binary_visualize(XY)

array([0.00669285, 0.01127661, 0.0189398 , 0.03164396, 0.05241435,
       0.08561266, 0.1368025 , 0.21151967, 0.31228169, 0.43458759,
       0.56541241, 0.68771831, 0.78848033, 0.8631975 , 0.91438734,
       0.94758565, 0.96835604, 0.9810602 , 0.98872339, 0.99330715],
      dtype=float128)

relu activation

In [6]:
X = np.linspace(-5,5,20)
Y = Activations.relu(X)
XY = np.stack([X, Y], axis=1)
print(repr(Y))
binary_visualize(XY)

array([ 0.00000000,  0.00000000,  0.00000000,  0.00000000,  0.00000000,
        0.00000000,  0.00000000,  0.00000000,  0.00000000,  0.00000000,
        0.26315789,  0.78947368,  1.31578947,  1.84210526,  2.36842105,
        2.89473684,  3.42105263,  3.94736842,  4.47368421,  5.00000000])

leaky_relu activation

In [7]:
X = np.linspace(-5,5,20)
Y = Activations.leaky_relu(X)
XY = np.stack([X, Y], axis=1)
print(repr(Y))
binary_visualize(XY)

array([-1.50000000, -1.34210526, -1.18421053, -1.02631579, -0.86842105,
       -0.71052632, -0.55263158, -0.39473684, -0.23684211, -0.07894737,
        0.26315789,  0.78947368,  1.31578947,  1.84210526,  2.36842105,
        2.89473684,  3.42105263,  3.94736842,  4.47368421,  5.00000000])

softmax activations

In [9]:
X = np.linspace(-5,5,20)
Y = Activations.softmax(X)
XY = np.stack([X, Y], axis=1)
print(repr(Y))
binary_visualize(XY)

array([0.00001858, 0.00003145, 0.00005323, 0.00009011, 0.00015252,
       0.00025817, 0.000437  , 0.00073971, 0.00125209, 0.00211939,
       0.00358746, 0.00607243, 0.01027872, 0.01739862, 0.02945038,
       0.04985021, 0.08438068, 0.14282987, 0.24176593, 0.40923346],
      dtype=float128)
In [ ]:
In [ ]: