Basic Neural Models

A robust implementation of fundamental neural network architectures, featuring an interactive Streamlit dashboard for training, visualization, and prediction.

Implemented Models

This repository contains implementations of:

• Single Layer Perceptron (SLP): The classic linear binary classifier.
• Adaptive Linear Neuron (ADALINE): Uses continuous predicted values for learning (Delta Rule).
• Multi-Layer Perceptron (MLP): A flexible feedforward neural network with Backpropagation, capable of solving non-linear multiclass classification problems.
  • Customizable Architecture: Configure the number of hidden layers and neurons per layer.
  • Activation Functions: Supports Sigmoid, Tanh, Linear, and Signum.
  • Modular Design: Built on reusable Layer and NeuralNetwork classes.

Features

• Interactive UI: Built with Streamlit for a seamless user experience.
• Dynamic Model Creation:
  • SLP/Adaline: Configure learning rates, epochs, bias, and thresholds.
  • MLP: Define custom network topology (hidden layers/neurons) and activation functions.
• Visualizations:
  • Decision Boundary plots.
  • Confusion Matrices.
  • Accuracy metrics (Train/Test).
  • Multiclass performance break-down.
• Prediction System: Load trained models and make real-time predictions on new data.
• Data Processing: Works with the Penguin data set, handling feature scaling and encoding automatically.

Getting Started

Prerequisites

• Python 3.8+
• Pip

Installation

1. Clone the repository:

   git clone <repository-url>
   cd Basic-Neural-Models

2. Install dependencies:

   pip install -r requirements.txt

Usage

Run the Streamlit application:

streamlit run main.py

Navigate through the sidebar to:

1. Create Model: Train and visualize SLP or Adaline models.
2. Predict: Use saved SLP/Adaline models for inference.
3. Back-Propagation: Design, train, and test custom MLP networks.