CNN Implementation Project

A convolutional neural network implementation from scratch for image classification using PyTorch.

Project Description

This project implements a CNN model for classifying images from the MNIST/CIFAR-10 dataset. The implementation focuses on understanding the fundamentals of convolutional neural networks, including:

• Building CNN architectures with convolutional, pooling, and fully-connected layers
• Training neural networks using backpropagation and gradient descent
• Evaluating model performance on image classification tasks

Setup Instructions

Environment Setup

1. Clone this repository: git clone https://github.com/sahith-M/cnn-project.git cd cnn-project

2. Create and activate a conda environment: conda create -n cnn_env python=3.9 conda activate cnn_env

3. Install required packages: conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia conda install jupyter matplotlib scikit-learn pandas

4. Run the project

Model Architecture

A simple CNN with:

• 2 convolutional layers
• 2 pooling layers
• 2 fully connected layers
• Output layer for classification

Implementation Timeline

• Data preprocessing and exploration
• Model architecture design
• Training implementation
• Evaluation and visualization
• Fine-tuning and optimization

Acknowledgments

Based on PyTorch tutorials and CS231n course materials