Neural Network from Scratch in Rust

This project implements a simple feedforward neural network in pure Rust, with no external machine learning libraries. It learns logic gates such as the CNOT using basic gradient descent and backpropagation.

---

Features :

• Built entirely from scratch using ndarray
• Implements:
  • Dense layers with sigmoid activation
  • Binary cross-entropy loss
  • Gradient descent optimizer
• Learns 2-input/2-output (CNOT) logic
• Trains using mini-batch (1-sample) updates

---

Project Structure :

Neural-network-from-scratch/
├── src/
│ └── main.rs # Main implementation
├── Cargo.toml # Dependencies and package config
├── .gitignore # Ignores build artifacts

---

Running the Project :

Make sure you have Rust and Cargo installed.

cargo run --release

This will compile and run the neural network training loop. You’ll see the training loss and final predictions.

---

Example Logic Gate: CNOT

CNOT gate truth table:

Input	Output
[1, 1, 0]	[1, 1, 1]
[1, 1, 1]	[1, 1, 0]

The model will learn to mimic this logic through training.