Fruit Classifier Neural Network

This repository contains a minimal implementation of a single-layer neural network built from scratch using only NumPy. It is designed as a learning tool to understand the core principles of how neural networks work under the hood.

Features and Concepts Covered

• Single-layer feedforward neural network (no hidden layers)
• Random weight initialization
• Forward propagation using the sigmoid activation function
• Backpropagation using the sigmoid derivative
• Gradient descent for weight optimization
• Binary classification using 2 input features
• Manual training loop over multiple iterations
• No machine learning libraries used (only NumPy)

The model is trained to classify fruit based on weight and color (0-green, 1-yellow) and output a binary classifier (0-lime, 1-lemon)

Requirements

• Python 3.x
• NumPy
• Understanding of
  • Sigmoid Functions
  • Derivatives
  • Gradients
  • Dot Product

License

MIT License. You are free to use, modify, and share the code.