Neural Network From Scratch

Implemented a Convolutional Neural Network (CNN) in C++ and a Feedforward Neural Network in C for MNIST digit classification, without using any external machine learning libraries.

Highlights:

• Full backpropagation implementation from scratch
• Matrix operations and optimization algorithms (SGD, momentum)
• Achieves competitive accuracy on MNIST dataset
• Demonstrates deep understanding of neural network internals

Technologies: C++, C, Mathematics, MNIST

GitHub:

• CNN in C++
• NN in C