README.md

miniMNIST-c

This project implements a mini neural network in C for classifying handwritten digits from the MNIST dataset.

Features

• Two-layer neural network (input -> hidden -> output)
• ReLU activation for hidden layer
• Softmax activation for output layer
• Cross-entropy loss function
• Stochastic Gradient Descent (SGD) with momentum
• L2 regularization

Prerequisites

• GCC compiler
• MNIST dataset files:
  • train-images.idx3-ubyte
  • train-labels.idx1-ubyte

Compilation

gcc -O3 -march=native -ffast-math -o nn nn.c -lm

Usage

1. Place the MNIST dataset files in the data/ directory.
2. Compile the program.
3. Run the executable:

./nn

The program will train the neural network on the MNIST dataset and output the accuracy and average loss for each epoch.

Configuration

You can adjust the following parameters in nn.c:

• HIDDEN_SIZE: Number of neurons in the hidden layer
• LEARNING_RATE: Initial learning rate
• MOMENTUM: Momentum coefficient for SGD
• L2_LAMBDA: L2 regularization coefficient
• EPOCHS: Number of training epochs
• BATCH_SIZE: Mini-batch size for training

License

This project is open-source and available under the MIT License.