This project implements a simple neural network from scratch using Python and NumPy. It includes a custom Value class for automatic differentiation and an MLP (Multi-Layer Perceptron) model for basic machine learning tasks.
Custom Computation Graph: The Value class enables forward and backward computations with support for operations like addition, multiplication, exponentiation, and activation functions (ReLU, tanh, exp).
Multi-Layer Perceptron (MLP): A simple feedforward neural network supporting training using gradient descent.
Backpropagation: The model computes gradients and updates weights using backpropagation.
├── Value.py # Custom class for handling computation graph and automatic differentiation
├── NeuralNetworks.py # Implements the MLP model
├── train.py # Training script
└── README.md # Project documentation