A learning project to understand neural networks by tackling the classic MNIST handwritten digit recognition problem using PyTorch.

Takes images of handwritten digits (0-9) and predicts which digit it is.

This project helped me learn:

• How to build and train neural networks with PyTorch
• Working with image data and CNN architectures
• Proper ML practices (validation, testing, metrics)
• Model evaluation and performance analysis

• Training/validation/test pipeline
• CNN model implementation
• Detailed performance metrics per digit
• Confusion matrix analysis

• Achieved high accuracy across all digit classes
• Consistent performance with <1% variance between digits
• Robust generalization on unseen test data

• Small TKinter Canvas drawing application which will predict drawn digits