Neural-Network

This project implements a simple feedforward neural network from scratch using Python and NumPy. It is designed for multi-class classification tasks such as classifying digits.

I have wanted to do this project since learning necessary concepts from multivariate calculus.

Features

• Two-layer neural network with ReLU activation and softmax output.
• Mini-batch gradient descent training.
• Cross-entropy loss function.
• One-hot encoded labels support.
• Training data shuffling and batching.
• Test accuracy evaluation.
• Visualization tool to display test images with predicted and true labels.

Usage

1. Train the model using the train function with your dataset.
2. Evaluate test accuracy.
3. Visualize predictions on test images with true and predicted labels.

Example Training Output

Epoch 1/10, Loss: 1.3596

Epoch 2/10, Loss: 0.5443

Epoch 3/10, Loss: 0.4162

Epoch 4/10, Loss: 0.3673

Epoch 5/10, Loss: 0.3394

Epoch 6/10, Loss: 0.3199

Epoch 7/10, Loss: 0.3044

Epoch 8/10, Loss: 0.2916

Epoch 9/10, Loss: 0.2805

Epoch 10/10, Loss: 0.2706

Test accuracy: 92.74%

Misclassified Images

Here is an example of a misclassified image: