====== Lectures ======

Please be aware that Mondays are largely affected by Public holidays. Some Mondays are thus substituted by other days. Lectures will be delivered by [[http://cmp.felk.cvut.cz/~svoboda|Tomas Svoboda]] unless stated otherwise.

PDF of the slides will be posted on this page. Note, however, that lectures will contain a significant portion of blackboard writing as well as some live programming pieces. Reading/watching slides only is not enough. Active participation in lectures is welcomed and encouraged.

^ date ^ week ^ topic ^
| 20.02.2017 |  1. | Intro into course and cybernetics and AI. We learn about the course grading and its rules as well as about what connects the cybernetics and artificial intelligence. I show some real applications of the methods we learn during the courses in an attempt to make the first lecture not too much dry and boring. {{:courses:be5b33kui:lectures:kui-01-intro.pdf|}}, {{:courses:be5b33kui:lectures:chapter01.pdf|}}|
| 27.02.2017 |  2. | Solving problems by search. We learn how to formalize problems and how to design //algorithms// for finding a solution. {{:courses:be5b33kui:lectures:chapter03.pdf|}}, {{:courses:be5b33kui:lectures:kui-02-search.pdf|}} (not completed, continue on March 13)  |
| 06.03.2017 |  3. | {{:courses:be5b33kui:lectures:kui-03-embodiedai.pdf|Embodied artificial intelligence}}, a lecture by [[https://sites.google.com/site/matejhof/home|Matej Hoffmann]] - a slight diversion into an interesting perspective on AI and robotics. |
| 13.03.2017 |  4. | Uniformed search, properties, comparisons. {{:courses:be5b33kui:lectures:kui-03-search.pdf|}}|
| 20.03.2017 |  5. | Informed search; {{:courses:be5b33kui:lectures:kui-04-informed-search.pdf|}} We find a way how to make search more effective by inserting some knowledge about the problem.  Adversarial search. How to find an optimal move/solution when someone or something is playing against us. |
| 27.03.2017 |  6. | Adversarial search. How to find an optimal move/solution when someone or something is playing against us. {{:courses:be5b33kui:lectures:chapter06.pdf|}}, {{:courses:be5b33kui:lectures:kui-05-adversarial.pdf|}}. Sequential decisions. What if an action (robot command, selling order) has an uncertain but probable outcome? How to compute the utilities? |
| 03.04.2017 |  7. | Sequential decisions. {{:courses:be5b33kui:lectures:kui-06-mdp.pdf|}} Value iteration, first compute state utilities than extract the best policy. |
| 10.04.2017 |  8. | Sequential decisions. Estimate the right policy directly - policy iteration method. {{:courses:be5b33kui:lectures:kui-07-mdp.pdf|}} Intro into the reinforcement learning. |
| <del>17.04.2017</del> |  9. ^ Public holiday - Easter Monday |
| 24.04.2017 |  10. | Reinforcement learning. What if nothing is known about the probability of action outcomes and we have to learn from final success or failure? {{:courses:be5b33kui:lectures:kui-08-rl.pdf|}} |
| <del>01.05.2017</del> |  11. ^ Public holiday |
| **02.05.2017** |  11. | Bayesian classification and decisions - [[https://sites.google.com/site/matejhof/home|Matej Hoffmann]]. How to decide optimally if we know all the (conditional) probabilities. {{:courses:be5b33kui:lectures:kui-11-bayes.pdf|}} |
| <del>08.05.2017</del> |  12. ^ Public holiday |
| **11.05.2017** |  12. | Classification - [[https://sites.google.com/site/matejhof/home|Matej Hoffmann]]. Perceptron, k-nn and relationship to Bayesian classifier - {{:courses:be5b33kui:lectures:kui-12-classification.pdf|}}   |
| 15.05.2017 |  13. | Learning probabilistic models - [[https://sites.google.com/site/matejhof/home|Matej Hoffmann]]. Slides from Jiri Matas: {{:courses:be5b33kui:lectures:Matas_pr_03_parameter_estimation_2016_10_17.pdf|}} - up to slide 15. |
| 22.05.2017 |  14. | Solving problems on paper/blackboard - fully interactive lecture; preparing for the exam a bit |