Artificial intelligence and expert systems 1000-2N00SID
1. Intelligent searching for problem solutions in the state space (greedy heuristics, algorithm A* etc.) and iterative search through the solution space (simulated annealing, evolutionary strategies etc.), with particular emphasis on optimization problems with constraints (constraint satisfaction problems).
2. Strategies in two-player games, minimax algorithm, alpha-beta pruning, MCTS (Monte Carlo Tree Search), games with incomplete information, discussing to what extent the above strategies can be implemented within computer game realizations and whether games is their only application area.
3. Logic-based knowledge and problem representation, the propositional calculus, satisfiability checking, the first-order logic, the resolution method, forward- and backward-chaining algorithms in knowledge bases (including heuristic methods applied to the realizations of backward-chaining), selected applications of logic-based techniques in the areas of planning (including reductions of the planning instances to the satisfiability problem), communication in multi-agent systems, as well as in advisory systems.
4. Relationships between machine learning and inductive learning, symbolic (e.g. deriving rules and decision trees from the data) and analytic (e.g. artificial neural networks) methods of machine learning, unsupervised (including data clustering and self-organizing maps), supervised and reinforcement learning, the tasks of machine learning understood as optimization problems (e.g. searching for the minimal decision trees by using heuristic methods, as well as learning artificial neural networks by basing on the iterative improvement techniques such as error back-propagation or evolutionary methods). Also the discussion that the areas of machine learning and artificial intelligence are not identical although they can be very helpful to each other (e.g. in games etc.).
5. Selected approaches to modeling and reasoning under uncertainty, including the foundations of probabilistic models (e.g. Naive Bayes, Bayesian networks, examples of using probabilities and information entropy in machine learning based on modern extensions of artificial neural networks), the theory and applications of fuzzy logic (including e.g. applications in robotics) with particular emphasis on heuristic extraction of fuzzy models from the data, theory and applications of rough sets (e.g. applications in data analysis), as well as selected elements of multi-valued logics, modal logics and temporal logics.
6. Open discussion about the current trends of research and applications of artificial intelligence in various practical domains, including the aspects of cooperation and interaction between humans and intelligent systems.
Main fields of studies for MISMaP
mathematics
Type of course
elective monographs
elective courses
Mode
Blended learning
Remote learning
Requirements
Prerequisites (description)
Course coordinators
Learning outcomes
The goal is to present the foundations and applications of selected methods of artificial intelligence and advisory / expert systems. The students will acquire fundamental skills related to complex problem solving by employing the discussed methods (K_W01, K_W03, K_W04, K_W05, K_W06, K_W09). They will also acquire useful knowledge with respect to using the discussed methods in the projects that they participate in, as well as in potential future projects corresponding to their research and professional interests (K_U01, K_U03, K_U08, K_U09, K_U35 K_U38). With regard to social skills, the students should be able to take into account the opinion of the others and actively participate in discussions (K_K01, K_K07). The detailed acquired skills are in relation to background knowledge and problem solving in the six main areas described in the course description.
Assessment criteria
The course comprises of the exercises (zero-one pass, no mark) and the exam (the mark). Successful passing of the exercises is the required and satisfactory condition to enter the exam. Passing of the exercises is truly zero-one, with no additional points or partial marks that can influence the final exam-based mark.
Important components for passing of the exercises is sufficiently regular presence during the classes, as well as passing (zero-one again) of the test comprising of several tasks related to the artificial-intelligence-based problem solving. Additional / alternative criterion may correspond to the homework tasks. Further details can be specified by persons who teach particular groups.
Attending lectures is not formally required although it may significantly help to master the material. The exam has a written form. It consists (like the test) of several tasks related to the artificial-intelligence-based problem solving. During the exam, the students can rely on their materials, but the tasks need to be solved individually. The additional exam can take a form of the oral exam.
Bibliography
1. Mariusz Flasiński: Introduction to Artificial Intelligence (Springer 2016)
2. Stuart Russell, Peter Norvig: Artificial Intelligence: A Modern Approach
3. George Luger: AI: Structures and Strategies for Complex Problem Solving
4. Tom Mitchell: Machine Learning
Additional information
Information on level of this course, year of study and semester when the course unit is delivered, types and amount of class hours - can be found in course structure diagrams of apropriate study programmes. This course is related to the following study programmes:
- Bachelor's degree, first cycle programme, Computer Science
- Master's degree, second cycle programme, Bioinformatics and Systems Biology
- Master's degree, second cycle programme, Computer Science
Additional information (registration calendar, class conductors, localization and schedules of classes), might be available in the USOSweb system: