This project implements a VGG11-based convolutional neural network (CNN) for classifying handwritten digits from the MNIST dataset. The model is trained with batch normalization, dropout, and data augmentation to improve generalization and robustness against image transformations.
- VGG11-inspired architecture with 8 convolutional layers and 3 fully connected layers.
- Batch normalization & dropout for improved stability and regularization.
- Data augmentation using horizontal/vertical flips and noise perturbations.
- Evaluation under transformations to test robustness against flipped and noisy images.
- Training visualization with accuracy and loss plots over epochs.