Neural Network Implementation

This repository contains Python code for a simple neural network implementation Neural_Network.ipnb along with a playground.ipnb notebook, which is a Python migration of the TensorFlow Playground.

Files

Neural_Network.ipnb

This Jupyter notebook contains the implementation of a neural network with various activation functions (Sigmoid, ReLU, Weighted ReLU, Tanh), a mean squared cost function, and perceptron and layer classes. It also includes a training monitor and an example usage section that demonstrates training the neural network for the XOR problem and plotting the decision boundary.

playground.ipnb

This Jupyter notebook is a Python migration of the TensorFlow Playground, providing an interactive environment to experiment with neural networks. It includes implementations for nodes, error functions, activation functions, regularization functions, links, and utilities for building and training neural networks.

Usage

Neural_Network.ipnb

1. Open the Neural_Network.ipnb notebook in a Jupyter environment.
2. Run the cells sequentially to load the neural network implementation, train the network for the XOR problem, and plot the decision boundary.

playground.ipnb

1. Open the playground.ipnb notebook in a Jupyter environment.
2. Explore the provided implementation of the neural network in the TensorFlow Playground style.
3. Experiment with different network architectures, activation functions, and training parameters interactively.

Dependencies

• Python 3.x
• NumPy
• Matplotlib

Make sure to install these dependencies before running the notebooks.

Feel free to modify the code, experiment with different configurations, and use the provided implementations for educational purposes or as a starting point for your projects.