Lectures are delivered by Petr Pošík or Tomas Svoboda (TS).
The course starts in regular contact teaching regime, but might change to distance teaching if needed.
We strongly suggest to attend the lectures personally. But we plan to make the lectures hybrid, such that you can take part also online. BigBlueButton will be used to deliver the lectures and for discussions. You will find the link to the course room in BRUTE, on Course ⇒ Conference rooms. BRUTE will also contain the lecture recordings in several formats:
You can also watch the MP4 recordings from the last year distance teaching at the BE5B33KUI YouTube channel.
PDF of the slides will be posted on this page. Preliminary version (e.g., last year's) before the lecture. Latest version, including teacher's notes shortly after the lecture. If you spot mistakes in the slides, please report them. Important corrections may be rewarded with bonus points. Active participation in lectures is welcomed and encouraged.
Books, on-line resources, specialization courses will be referenced throughout the lectures - their list is typically at the last slide.
Date | Week | Content | Alternative lecture video from AI@Berkeley |
---|---|---|---|
14.02.2022 BBB recording | 1 | Rules of the game (grading, assignments, etc.). Cybernetics and AI - motivation. Course overview. Is every problem solvable? N-puzzle. 01_intro.pdf | |
21.02.2022 BBB recording | 2 | Solving problems by search. Trees and graphs. Completeness, Optimality, Complexity. DFS, BFS. 02_search.pdf, 02_search_handout_notes.pdf | Uninformed search |
28.01.2022 BBB recording | 3 | Solving problems by search. How to avoid looping forever and how to go faster to the goal. Informed search. Heuristics. A*. 03_search.pdf 03_search_live_withnotes.pdf | Informed search |
07.03.2022 Lecture not recorded. Use BBB recording from previous year. | 4 | Two player-games. Adversarial search - Search when playing against a (rational) opponent. 04_adversarial.pdf 04_adversarial_live_withnotes.pdf. | Adversarial search |
14.03.2022 BBB recording | 5 | Games with random elements, multi-player games. Expectimax. Utilities. 05_expectimax.pdf 05_expectimax_live_withnotes.pdf | Uncertainty and utilities |
21.03.2022 BBB recording | 6 | Decision-making under uncertainty I. Route to goal when action outcome is probabilistic. Value iteration. 06_mdp.pdf 06_mdp_live_withnotes.pdf | Markov Decision Processes |
28.03.2022 BBB recording | 7 | Decision-making under uncertainty II. Policy iteration. 07_mdp.pdf 07_mdp_live_withnotes.pdf | Markov Decision Processes II |
04.04.2022 BBB recording | 8 | Reinforcement learning I. What if nothing is known about the probability of action outcomes and we have to learn from final success or failure? 08_rl.pdf 08_rl_live_withnotes.pdf | Reinforcement learning |
11.04.2022 BBB recording | 9 | Reinforcement learning II. Exploration vs. exploitation. 09_rl.pdf 09_rl_live_withnotes.pdf | Reinforcement learning II |
18.04.2022 | Holiday - Easter Monday | ||
25.04.2022 | 10 | Mid-term exam from topics covered up to now. Inspiration: quizzes. | |
02.05.2022 BBB recording | 11 | Basic concepts of probability. Bayesian classification and decisions. 10_bayes.pdf 10_bayes_live_withnotes.pdf | Probability |
09.05.2022 BBB recording | 12 | Naive Bayesian classification, Laplace smoothing, Precision, Recall and ROC curve. 11_recog_a.pdf 11_recog_a_live_withnotes.pdf | Naive Bayes, ROC curve, video by Andrew Ng |
16.05.2022 BBB recording | 13 | Linear classifiers, perceptron. 11_recog_b.pdf 11_recog_b_live_withnotes.pdf | Perceptrons and Logistic Regression - the perceptron part, or Nearest neighbor (k-nn) classification, watch Kernels and Clustering - the nearest neighbors part |