Interactive Simple Artificial Neural Network (ANN)

Author: Harshitha Manjunatha

Overview

This project provides an interactive dashboard built from scratch to demonstrate the inner workings of a simple Artificial Neural Network (ANN). It visualizes parameter initialization, forward propagation, and backpropagation in real-time.

The interactive elements allow users to manipulate:

• Epochs ($n_{epoch}$): The number of training cycles.
• Learning Rate ($\eta$): The step size for gradient descent.
• Seed ($S$): The random number generator seed for reproducibility.
• Visualization Layers: Toggles for viewing Weights, Forward Pass, and Backpropagation derivatives.

Features

• Custom ANN Implementation: Coded purely in Python using NumPy (no high-level DL frameworks).
• Interactive Widgets: Powered by ipywidgets for real-time parameter tuning.
• Visualization: Custom matplotlib plotting to visualize nodes, connections, signal flow, and gradients.

Installation

1. Clone this repository:

   git clone [https://github.com/your-username/interactive_ann.git](https://github.com/your-username/interactive_ann.git)