This exercise gives an implementation of a perceptron.
Perceptron
- A perceptron has only input layers and output layers.
- No hidden layers.
- It uses step function as activation function.
- Perceptron output is +1 or -1. - Perceptron Learning rule is used to update the weights and bias.
Methodology:
The general methodology to build a Neural Network is to: 1. Define the neural network structure ( # of input units, # of hidden units, etc). 2. Initialize the model's parameters 3. Loop: - Implement forward propagation - Compute loss - Update parameters (Perceptron Learning rule)