Scribd
Upload
12 views5 pages

Neural Network Code

code

Uploaded by

kishoranbu14
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
12 views5 pages

Neural Network Code

code

Uploaded by

kishoranbu14
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 5

Neural Network Binary Classification Code

import numpy as np # For matrix math


import matplotlib.pyplot as plt # For plotting

# Training data and labels


X = np.array([
[0, 1],
[1, 0],
[1, 1],
[0, 0]
])
y = np.array([
[1],
[1],
[0],
[0]
])

# Parameters
num_i_units = 2 # Number of Input units
num_h_units = 2 # Number of Hidden units
num_o_units = 1 # Number of Output units
learning_rate = 0.01 # Learning rate
reg_param = 0 # Regularization parameter
max_iter = 5000 # Maximum iterations
m = len(X) # Number of training examples

# Initialize weights and biases


np.random.seed(1)
W1 = np.random.normal(0, 1, (num_h_units, num_i_units)) # 2x2
W2 = np.random.normal(0, 1, (num_o_units, num_h_units)) # 1x2
B1 = np.random.random((num_h_units, 1)) # 2x1
B2 = np.random.random((num_o_units, 1)) # 1x1

# Activation function
def sigmoid(z, derv=False):
if derv:
return sigmoid(z) * (1 - sigmoid(z))
return 1 / (1 + np.exp(-z))
# Forward propagation
def forward(x, predict=False):
a1 = x.reshape(x.shape[0], 1) # Convert input to column vector
z2 = W1.dot(a1) + B1 # 2x1
a2 = sigmoid(z2) # Hidden layer activation
z3 = W2.dot(a2) + B2 # 1x1
a3 = sigmoid(z3) # Output layer activation
if predict:
return a3
return a1, a2, a3

# Training function
def train(W1, W2, B1, B2):
cost_history = np.zeros((max_iter, 1)) # To store cost at each iteration

for i in range(max_iter):
c=0
dW1 = np.zeros_like(W1)
dW2 = np.zeros_like(W2)
dB1 = np.zeros_like(B1)
dB2 = np.zeros_like(B2)

for j in range(m):
# Forward propagation
a0 = X[j].reshape(X[j].shape[0], 1) # 2x1
z1 = W1.dot(a0) + B1 # 2x1
a1 = sigmoid(z1) # 2x1
z2 = W2.dot(a1) + B2 # 1x1
a2 = sigmoid(z2) # 1x1

# Backpropagation
dz2 = a2 - y[j] # 1x1
dW2 += dz2 * a1.T # 1x2
dB2 += dz2 # 1x1
dz1 = np.multiply(W2.T.dot(dz2), sigmoid(z1, derv=True)) # 2x1
dW1 += dz1.dot(a0.T) # 2x2
dB1 += dz1 # 2x1

# Compute cost
c += -y[j] * np.log(a2) - (1 - y[j]) * np.log(1 - a2)

# Update weights and biases


W1 -= learning_rate * (dW1 / m) + (reg_param / m) * W1
W2 -= learning_rate * (dW2 / m) + (reg_param / m) * W2
B1 -= learning_rate * (dB1 / m)
B2 -= learning_rate * (dB2 / m)

# Store the average cost


cost_history[i] = c / m

return W1, W2, B1, B2, cost_history

# Train the network


W1, W2, B1, B2, cost_history = train(W1, W2, B1, B2)

# Plot cost over iterations


plt.plot(range(max_iter), cost_history)
plt.xlabel("Iterations")
plt.ylabel("Cost")
plt.title("Cost Reduction Over Iterations")
plt.show()

import numpy as np # For matrix math


import matplotlib.pyplot as plt # For plotting

# Training data and labels


X = np.array([
[0, 1],
[1, 0],
[1, 1],
[0, 0]
])
y = np.array([
[1],
[1],
[0],
[0]
])

# Parameters
num_i_units = 2 # Number of Input units
num_h_units = 2 # Number of Hidden units
num_o_units = 1 # Number of Output units
learning_rate = 0.01 # Learning rate
reg_param = 0 # Regularization parameter
max_iter = 5000 # Maximum iterations
m = len(X) # Number of training examples

# Initialize weights and biases


np.random.seed(1)
W1 = np.random.normal(0, 1, (num_h_units, num_i_units)) # 2x2
W2 = np.random.normal(0, 1, (num_o_units, num_h_units)) # 1x2
B1 = np.random.random((num_h_units, 1)) # 2x1
B2 = np.random.random((num_o_units, 1)) # 1x1

# Activation function
def sigmoid(z, derv=False):
if derv:
return sigmoid(z) * (1 - sigmoid(z))
return 1 / (1 + np.exp(-z))

# Forward propagation
def forward(x, predict=False):
a1 = x.reshape(x.shape[0], 1) # Convert input to column vector
z2 = W1.dot(a1) + B1 # 2x1
a2 = sigmoid(z2) # Hidden layer activation
z3 = W2.dot(a2) + B2 # 1x1
a3 = sigmoid(z3) # Output layer activation
if predict:
return a3
return a1, a2, a3

# Training function
def train(W1, W2, B1, B2):
cost_history = np.zeros((max_iter, 1)) # To store cost at each iteration

for i in range(max_iter):
c=0
dW1 = np.zeros_like(W1)
dW2 = np.zeros_like(W2)
dB1 = np.zeros_like(B1)
dB2 = np.zeros_like(B2)

for j in range(m):
# Forward propagation
a0 = X[j].reshape(X[j].shape[0], 1) # 2x1
z1 = W1.dot(a0) + B1 # 2x1
a1 = sigmoid(z1) # 2x1
z2 = W2.dot(a1) + B2 # 1x1
a2 = sigmoid(z2) # 1x1

# Backpropagation
dz2 = a2 - y[j] # 1x1
dW2 += dz2 * a1.T # 1x2
dB2 += dz2 # 1x1
dz1 = np.multiply(W2.T.dot(dz2), sigmoid(z1, derv=True)) # 2x1
dW1 += dz1.dot(a0.T) # 2x2
dB1 += dz1 # 2x1

# Compute cost
c += -y[j] * np.log(a2) - (1 - y[j]) * np.log(1 - a2)

# Update weights and biases


W1 -= learning_rate * (dW1 / m) + (reg_param / m) * W1
W2 -= learning_rate * (dW2 / m) + (reg_param / m) * W2
B1 -= learning_rate * (dB1 / m)
B2 -= learning_rate * (dB2 / m)

# Store the average cost


cost_history[i] = c / m

return W1, W2, B1, B2, cost_history

# Train the network


W1, W2, B1, B2, cost_history = train(W1, W2, B1, B2)

# Plot cost over iterations


plt.plot(range(max_iter), cost_history)
plt.xlabel("Iterations")
plt.ylabel("Cost")
plt.title("Cost Reduction Over Iterations")
plt.show()

You might also like