3rd Edition
Deep and Reinforcement Learning
Barcelona
UPC ETSETB TelecomBCN (Autumn 2020)
This course presents the principles of reinforcement learning as an artificial intelligence tool based on the interaction of the machine with its environment, with applications to control tasks (eg. robotics, autonomous driving) o decision making (eg. resource optimization in wireless communication networks). It also advances in the development of deep neural networks trained with little or no supervision, both for discriminative and generative tasks, with special attention on multimedia applications (vision, language and speech).
Instructors
RL
Reinforcement Learning
- (JV) Introduction to reinforcement learning
- (JV) Armed Bandits
- (MC) Example: Introduction to the gridworld
- (JV) Markov Decision Processes. Bellman Equations,
- (MC) Example: Recycling robot
- (JV) Dynamic programming
- (MC) Example: Jack’s car rental
- (JV) Monte Carlo methods
- (MC) Example: Blackjack
- (JV) Time difference methods
- (VC) Lab: Tabular Q-Learning
- (JV) Value function approximation
- (MC) Demo: Gridworld over a cliff
- (JV) Policy Gradient
- (JV) Actor-Critic
Labs
- (ALL) MULTI ARMED BANDITS
- (ALL) DYNAMIC PROGRAMMING
- (ALL) TEMPORAL DIFFERENCE LEARNING
(ALL): MC & JV & JN
DRL
Deep Reinforcement Learning
- (XG) Deep Q-Networks [GSlides@upc.edu]
- (XG) Deep Policy Gradient [GSlides@upc.edu]
- (XG) Deep Actor-Critic [GSlides@upc.edu]
- (VC) Towards RL that scales
Labs
- (JN) Lab: Q-Learning with Neural Networks
- (JN) Lab: REINFORCE & REINFORCE w/ baseline
ADL
Advanced Deep Learning (ADL)
Check our Deep Learning teaching repository for introductory contents in DL.
- (XG) Transfer Learning
- (XG) Generative Adversarial Networks (GANs)
- (JT) Supercomputing Talk
- (AM) Graph Neural Newtorks (GNN)
- (FA) Meta-Learning Talk
- (XG) Attention-based Models
- (IC) RL for Real-world Robotics Talk
- (OV) From AlphaGo to AlphaStar Talk I Talk II
- (CF) RL: What supervision scales? Talk
Labs
- (AM) Transfer Learning in vision
- (AM) GANs
- (AM) GNN for Recommender Systems
- (AM) RNN & Transformer
Practical
Practical details
- Study Programs: Bachelor degrees on Data Science Engineering (GCED) and Telecommunication Engineerning (GRETST at ETSETB TelecomBCN from the Universitat Politecnica de Catalunya in Barcelona.
- Course code and official guide: 230817 - ARAP
- Requirements: Aprenentatge Automatic 2 (GCED) or Introduction to Deep Learning (GRETST). MSc students cannot sign up to this course.
- Language: Catalan (course material in English)
- ECTS credits: 6 ECTS
- Semester: Autumn 2020
- Dates: September 14 - December 14, 2020
- Schedule: Lectures on Mondays 12pm-2pm, Labs on Mondays 10am-12pm or Tuesdays 12pm-2pm
- Room: A4104+A4105 (Lectures) / A4105 (Labs)
- Location: Campus Nord UPC Barcelona
- Course on Atenea
- Feedback & Questions: Please use the [Github Issues] section.(https://github.com/telecombcn-dl/drl-2020/issues).
Organizers
ETSETB Telecom BCN
Universitat Politècnica de Catalunya