Learning Outcomes
Upon successful completiion of this course, students will have a grasp of the fundamentals around the theory and practical application of modern deep learning models. These will allow students to apply machine / deep learning techniques in a production environment, as well as to carry out research in relevant fields.
Course Content
- Representation learning for signals - convolutional neural networks (CNN)
- Capacity control: Regularization, dropout and data augmentation
- Popular CNN architectures
- Word embeddings
- Data loading, preprocessing and training workflows in pytorch
- Recurrent neural networks (RNN) with emphasis on LSTM and GRU
- Attention mechanisms
- Transformer networks
- Generative models: GAN, VAE and Normalizing Flows
- Introduction to Reinforcement Learning
- DQN, A2C and PPO algorithms
