Advanced Neural Network with Live Visualizations 🚀

Overview

This project implements an Advanced Convolutional Neural Network (CNN) for image classification using the CIFAR-10 dataset. The model features real-time loss and accuracy tracking with dynamic visualizations during training, making it an exciting way to see deep learning in action.

Features

✅ Advanced CNN Architecture with multiple convolutional layers
✅ Live Training Progress Visualization for loss and accuracy
✅ Simulated Feature Map Display for real-time input image visualization
✅ Runs on Google Colab with GPU acceleration support

Setup & Execution

1. Open Google Colab and upload the notebook or run the script in Python.
2. Install dependencies (if needed):

   !pip install torch torchvision matplotlib

3. Train the model and watch live graphs updating dynamically!

Example Output

During training, the following real-time graphs are displayed: 📊 Loss Graph (Shows how the model improves)
📈 Accuracy Graph (Tracks training performance)
🖼 Live Input Image (Displays real-time training images with class labels)

Model Architecture

• 3 Convolutional Layers with ReLU activation
• Max Pooling Layers for downsampling
• Fully Connected Layers for classification
• Dropout Layer for regularization
• Adam Optimizer with CrossEntropy Loss

Results

Achieved ~80% test accuracy on the CIFAR-10 dataset.

How to Run the Code

• On Google Colab: Upload the script or run the notebook cells sequentially.
• Locally: Ensure torch, torchvision, and matplotlib are installed, then execute the script.

Author

Created by Paras Singh. Connect with me on LinkedIn! 🚀