Computer Science

1. Introduction to Artificial Intelligence (AI), history and areas of AI 2. Agents, rationality, task environments 3. Non-informed search strategies: breadth-first, min-cost, depth-first, bounded depth, iterative deepening 4. Informed search strategies: best-first greedy, A*, heuristics: admissibility and consistency 5. Iterative refinement algorithms: hill-climbing, steepest-descent, simulated annealing, local beam, genetic algorithms 6. Constraint Satisfaction Problems: definition, variants, constraint propagation (node-, (generalised-)arc-, path-, K-consistency), backtracking, backjumping, local search 7. Knowledge and reasoning, logic-based agents, propositional logic, SAT 8. Knowledge and reasoning in first-order logic 9. Classical planning 10. Knowledge and reasoning under uncertainty: Bayesian networks, electives in decision theory 11. Electives in machine-learning: induction of decision trees; evaluation of performance of a learning algorithm, loss functions; model selection; ensemble learning: bagging, random forests, boosting; explainability, adversarial examples; parametric vs. non-parametric learning; k-nearest neighbours.

Students are expected to have good knowledge of: 1) Imperative and object-oriented programming in any high-level language 2) Design of algorithms and data structures 3) General mathematics, algebra, set theory, probability theory 4) Propositional and first-order logic 5) Design of relational databases 6) Automata, computational complexity, decidability.

Slides and lecture notes made available by the teacher.

The course comprises lecture classes and labs covering the whole course programme.

Attendance is not mandatory, but highly recommended.

The exams comprises two written parts (Prova A and Prova B) and a student project. In the written exams, students are requested to answer to questions on the entire programme. The two written exams cover half the course programme each (Part A and Part B, respectively). The student project consists in a written report (plus, possibly, software) to be presented in a seminar. Each written exam will contribute to the 30% of the final score. The student project and its presentation in a semina will contribute to the remaining 40%.

Further readings: Stuart Russell, Peter Norvig Artificial Intelligence, a Modern Approach. 4th Edition Pearson. ISBN 978-1292401133.

The course comprises lectures and labs covering the whole course programme.