This project implements a neural network from scratch to solve the XOR problem. The network demonstrates the capability to learn with and without momentum, offering flexibility in training approaches. Features
Custom neural network implementation XOR problem solution Momentum-based learning option Choice of activation functions:
- Hyperbolic Tangent (tanh)
- Rectified Linear Unit (ReLU)
- Sigmoid
The XOR problem is a classic benchmark in neural network training, as it requires a multi-layer network to solve. This implementation showcases how a simple neural network can learn to correctly classify XOR inputs.
- Python