Lokizzer/Experiment-5-Implementation-of-XOR-using-RBF

Experiment-5-Implementation-of-XOR-using-RBF

AIM:

To classify the Binary input patterns of XOR data by implementing Radial Basis Function Neural Networks.

EQUIPMENTS REQUIRED:

Hardware – PCs Anaconda – Python 3.7 Installation / Google Colab /Jupiter Notebook

RELATED THEORETICAL CONCEPT:

Exclusive or is a logical operation that outputs true when the inputs differ.For the XOR gate, the TRUTH table will be as follows XOR truth table

XOR is a classification problem, as it renders binary distinct outputs. If we plot the INPUTS vs OUTPUTS for the XOR gate, as shown in figure below

The graph plots the two inputs corresponding to their output. Visualizing this plot, we can see that it is impossible to separate the different outputs (1 and 0) using a linear equation.

A Radial Basis Function Network (RBFN) is a particular type of neural network. The RBFN approach is more intuitive than MLP. An RBFN performs classification by measuring the input’s similarity to examples from the training set. Each RBFN neuron stores a “prototype”, which is just one of the examples from the training set. When we want to classify a new input, each neuron computes the Euclidean distance between the input and its prototype. Thus, if the input more closely resembles the class A prototypes than the class B prototypes, it is classified as class A ,else class B.

A Neural network with input layer, one hidden layer with Radial Basis function and a single node output layer (as shown in figure below) will be able to classify the binary data according to XOR output.

The RBF of hidden neuron as gaussian function

ALGORIHM:

STEP 1: Import the required Python libraries

STEP 2: Define Activation Function : Sigmoid Function

STEP 3: Initialize neural network parameters (weights, bias) and define model hyperparameters (number of iterations, learning rate)

STEP 4: Forward Propagation and Backward Propagation

STEP 5: Test the model

PROGRAM:

import numpy as np
import matplotlib.pyplot as plt
import tensorflow as tf
from tensorflow.keras.initializers import Initializer
from tensorflow.keras.layers import Layer
from tensorflow.keras.initializers import RandomUniform, Initializer, Constant

def gaussian_rbf(x, landmark, gamma=1):
    return np.exp(-gamma * np.linalg.norm(x - landmark)**2)
 def end_to_end(X1, X2, ys, mu1, mu2):
    from_1 = [gaussian_rbf(i, mu1) for i in zip(X1, X2)]
    from_2 = [gaussian_rbf(i, mu2) for i in zip(X1, X2)]
    # plot
    
    plt.figure(figsize=(13, 5))
    plt.subplot(1, 2, 1)
    plt.scatter((x1[0], x1[3]), (x2[0], x2[3]), label="Class_0")
    plt.scatter((x1[1], x1[2]), (x2[1], x2[2]), label="Class_1")
    plt.xlabel("$X1$", fontsize=15)
    plt.ylabel("$X2$", fontsize=15)
    plt.title("Xor: Linearly Inseparable", fontsize=15)
    plt.legend()

    plt.subplot(1, 2, 2)
    plt.scatter(from_1[0], from_2[0], label="Class_0")
    plt.scatter(from_1[1], from_2[1], label="Class_1")
    plt.scatter(from_1[2], from_2[2], label="Class_1")
    plt.scatter(from_1[3], from_2[3], label="Class_0")
    plt.plot([0, 0.95], [0.95, 0], "k--")
    plt.annotate("Seperating hyperplane", xy=(0.4, 0.55), xytext=(0.55, 0.66),
                arrowprops=dict(facecolor='black', shrink=0.05))
    plt.xlabel(f"$mu1$: {(mu1)}", fontsize=15)
    plt.ylabel(f"$mu2$: {(mu2)}", fontsize=15)
    plt.title("Transformed Inputs: Linearly Seperable", fontsize=15)
    plt.legend()

    # solving problem using matrices form
    # AW = Y
    A = []

    for i, j in zip(from_1, from_2):
        temp = []
        temp.append(i)
        temp.append(j)
        temp.append(1)
        A.append(temp)
    
    A = np.array(A)
    W = np.linalg.inv(A.T.dot(A)).dot(A.T).dot(ys)
    print(np.round(A.dot(W)))
    print(ys)
    print(f"Weights: {W}")
    return W
def predict_matrix(point, weights):
    gaussian_rbf_0 = gaussian_rbf(np.array(point), mu1)
    gaussian_rbf_1 = gaussian_rbf(np.array(point), mu2)
    A = np.array([gaussian_rbf_0, gaussian_rbf_1, 1])
    return np.round(A.dot(weights))
x1 = np.array([0, 0, 1, 1])
x2 = np.array([0, 1, 0, 1])
ys = np.array([0, 1, 1, 0])

# centers
mu1 = np.array([0, 1])
mu2 = np.array([1, 0])

w = end_to_end(x1, x2, ys, mu1, mu2)

# testing

print(f"Input:{np.array([0, 0])}, Predicted: {predict_matrix(np.array([0, 0]), w)}")
print(f"Input:{np.array([0, 1])}, Predicted: {predict_matrix(np.array([0, 1]), w)}")
print(f"Input:{np.array([1, 0])}, Predicted: {predict_matrix(np.array([1, 0]), w)}")
print(f"Input:{np.array([1, 1])}, Predicted: {predict_matrix(np.array([1, 1]), w)}")

mu1 = np.array([0, 0])
mu2 = np.array([1, 1])

w = end_to_end(x1, x2, ys, mu1, mu2)

# testing

print(f"Input:{np.array([0, 0])}, Predicted: {predict_matrix(np.array([0, 0]), w)}")
print(f"Input:{np.array([0, 1])}, Predicted: {predict_matrix(np.array([0, 1]), w)}")
print(f"Input:{np.array([1, 0])}, Predicted: {predict_matrix(np.array([1, 0]), w)}")
print(f"Input:{np.array([1, 1])}, Predicted: {predict_matrix(np.array([1, 1]), w)}")

OUTPUT :

RESULT:

Thus, the implementaion of XOR using RBF is executed successfully.