A foundational exploration into deep learning and machine vision. I engineered a neural network architecture capable of ingesting raw pixel data, processing it through hidden layers, and accurately predicting the corresponding integer.
The Architecture
Built to understand the core mathematics behind neural networks, backpropagation, and gradient descent. The model processes 28x28 pixel grids and maps them to probability distributions.
Tech Stack
- Python
- TensorFlow / PyTorch
- NumPy & Matplotlib for data visualization