🧠 Neural Network From Scratch in Python

This repository contains the full implementation of a simple neural network built from scratch in Python, as demonstrated in my Medium article.

---

📖 Overview

In this project, we build a fully connected neural network from scratch without using deep learning libraries like PyTorch or TensorFlow. The implementation includes:

• Forward Propagation
• Backpropagation
• Training with Gradient Descent
• Evaluation & Accuracy Calculation

The code is designed to be simple and educational, demonstrating the core concepts of neural networks. Perfect for beginners who want to understand how neural networks work under the hood!

---

📂 Files

• simple_nn.py — Implementation of the neural network.
• train.py — Training script with evaluation functions.
• dataset.py — Helper functions for data processing and evaluation.

---

💡 Getting Started

Requirements

• Python 3.x
• NumPy

Install dependencies:

pip install numpy
pip install matplotlib

Usage:

%matplotlib inline
from dataset import visualize_classification, XORDataset
import matplotlib.pyplot as plt
from simple_nn import SimpleClassifier, GradientDescent
from train_nn import train_model, eval_model, create_data_loader

num_inputs = 2
num_hidden = 4
num_outputs = 1

train_dataset = XORDataset(size=2500)
test_dataset = XORDataset(size=500)

train_data_loader = create_data_loader(train_dataset)
test_data_loader = create_data_loader(test_dataset)


model = SimpleClassifier(num_inputs, num_hidden, num_outputs)
optimizer = GradientDescent(lr=0.01)

_ = visualize_classification(model, test_dataset.data, test_dataset.label)

train_model(model, train_data_loader, optimizer)
eval_model(model, test_data_loader)
_ = visualize_classification(model, test_dataset.data, test_dataset.label)

Model Accuracy: 100.00%