Neural Networks and Deep Learning Course - University of Tehran

This repository contains the homework assignments for the Neural Networks and Deep Learning course at the University of Tehran. The assignments cover key concepts like CNNs, U-Net, RNNs, VAE, GANs, and Transformers, focusing on tasks such as image classification, semantic segmentation, object detection, and generative modeling.

Homeworks Overview

Homework 1:

• Topics: Introduction to CNNs, backpropagation, feed-forward networks.
• Key Problems:
  • Implementing a basic convolutional neural network.
  • Understanding the backpropagation algorithm.

Homework 2:

• Topics: Alzheimer's disease classification using CNNs, data augmentation.
• Key Problems:
  • Implementing a CNN to classify Alzheimer’s disease from brain MRI images.
  • Data augmentation techniques to improve model generalization.

Homework 3:

• Topics: U-Net architecture, semantic segmentation.
• Key Problems:
  • Implementing U-Net for MRI image segmentation.
  • Data augmentation and evaluation using Dice Coefficient and IoU.

Homework 4:

• Topics: Faster R-CNN for object detection.
• Key Problems:
  • Implementing Faster R-CNN for underwater object detection.
  • Comparison of region-based CNN models (R-CNN, Fast R-CNN, and Faster R-CNN).

Homework 5:

• Topics: Transformers in image classification.
• Key Problems:
  • Fine-tuning Vision Transformers (ViT) for CIFAR-10 image classification.
  • Comparison with CNN-based models.

Homework 6:

• Topics: Variational Autoencoders (VAE) and image generation.
• Key Problems:
  • Implementing VAE and Conditional VAE for anime and cartoon face datasets.
  • Exploring VQ-VAE models and analyzing generated outputs.