Last active
July 25, 2019 14:29
-
-
Save Jeraldy/b5bb83ed10df20834c75bab6b963bebd to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
""" | |
Jeraldy Deus | deusjeraldy@gmail.com | |
Implementing an Artificial Neural Network in numpy | |
BSD License | |
""" | |
import numpy as np | |
X = np.array([ | |
[0,0], | |
[0,1], | |
[1,0], | |
[1,1] | |
]) | |
Y = np.array([ | |
[0], | |
[1], | |
[1], | |
[0] | |
]) | |
m = X.shape[0] | |
num_nodes = 400 | |
W1 = np.random.randn(num_nodes,X.shape[1]) | |
b1 = np.zeros((num_nodes,1)) | |
W2 = np.random.randn(1,num_nodes) | |
b2 = np.zeros((1,X.shape[0])) | |
X = X.T | |
Y = Y.T | |
costs = [] | |
for i in range(4000): | |
# Foward Prop | |
# LAYER 1 | |
Z1 = np.dot(W1,X) + b1 | |
A1 = 1/(1+np.exp(-Z1)) | |
# LAYER 2 | |
Z2 = np.dot(W2,A1) + b2 | |
A2 = 1/(1+np.exp(-Z2)) | |
# Back Prop | |
dZ2 = A2 - Y | |
dW2 = (1/m)*np.dot(dZ2,A1.T) | |
db2 = (1/m)*np.sum(dZ2,axis=1,keepdims=True) | |
dZ1 = np.multiply(np.dot(W2.T, dZ2), 1 - np.power(A1, 2)) | |
dW1 = (1/m)*np.dot(dZ1,X.T) | |
db1 = (1/m)*np.sum(dZ1,axis=1,keepdims=True) | |
# Gradient Descent | |
W2 = W2 - 0.01*dW2 | |
b2 = b2 - 0.01*db2 | |
W1 = W1 - 0.01*dW1 | |
b1 = b1 - 0.01*db1 | |
# Loss | |
L = (-1/m)*np.sum(Y*np.log(A2) + (1-Y)*np.log(1-A2)) | |
L = np.squeeze(L) | |
if i%500 == 0: | |
print("=======================================") | |
print("Loss = ",L) | |
print("Predictions == ",A2) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment