WUT Deep Learning Course Repository

This repository contains all code I have developed during my the Deep Learning course at Warsaw University of Technology (official course name: "Artificial Neural Networks" - Sztuczne Sieci Neuronowe).

The objective of the course was to familiarize students with more advanced concepts from neural networks, i.e. their structures, methods of learning and potential applications. During the course, numerous neural network architectures have been presented, including multi-layer perceptron, convolutional and generative networks, auto-encoders, recurrent networks and others. Learning algorithms such as CM, NAG, Adam and others are also introduced.

Seminars and graded projects related to the subject were aimed at implementing the methods learned during the lectures and their use in solving practical problems. Additionally, the exercises cover numerous technical aspects and discuss good practices in using neural networks.

Course structure

The course comprised:

• 15 lectures - 30h
• 15 seminars - 30h
• 6 graded projects - individual work at home

All code has been written in Python combined with libraries such as PyTorch, TorchVision, scikit-learn, OpenCV and HuggingFace Transformers.

Concepts covered during Seminars and Projects:

• Regression and Classification
• Neural Networks
• Training, Optimization, Hyperparameter tuning
• Data Processing, Cleaning and Manipulation
• Image Processing, Transformation and Augmentation
• Convolutional Neural Networks (CNN)
• Recurrent Neural Networks (RNN)
• Natural Language Processing (NLP)
• Transformers
• AutoEncoders
• Generative Models (VAE, GAN etc.)
• Utilizing advanced Deep Learning libraries such as PyTorch Lightning, HuggingFace