LEARNING OUTCOMES

The goal of the course is to provide the foundations of knowledge-based intelligent autonomous agents.

AIMS AND LEARNING OUTCOMES

The course introduces the languages and the techniques through which intelligent agents can operate autonomously on a deductive basis. The aim of the course is to provide students with the capability of formalizing domains of interest in order to treat them in the context of autonomous intelligent agents, with specific reference to propositional logics, first-order logics, description logics, and automated planning languages. The main result is the student's ability to frame the problems in a formal way and abstract their main features in a specification which makes computationally feasible to implement autonomous agents.

PREREQUISITES

Some preliminary knowledge of combinatorics, algebra and theoretical computer science are useful for a better understanding of the course material.

TEACHING METHODS

Recorded lectures for the theory part;  practical sessions with the teacher to solve assigned exercises (possibly online)

SYLLABUS/CONTENT

Logical Agents: knowledge-based agents, logic, propositional logic (syntax, semantics, propositional knowledge bases, inference procedures, normal forms, resolution).

First-Order Logic: representation, syntax and semantics, knowledge engineering.

Inference in First-Order Logic: propositional vs. first-order inference, unification and lifting, resolution.

Knowledge representation: ontological engineering, categories and objects, events, reasoning systems for categories (semantic networks, description logics), reasoning with default information).

Classical Planning: definition, PDDL language, examples, planning as state-space search, planning graphs.

Planning in the real world: time, schedules and resources; hierarchical planning; planning and acting in nondeterministic domains.

Introduction to Reinforcement Learning: introduction, multi-armed bandits, finite markov decision processes, dynamic programming, Monte Carlo methods, Temporal-Difference learning, Planning and learning.

RECOMMENDED READING/BIBLIOGRAPHY

Stuart Russell, Peter Norvig - Artificial Intelligence, a Modern Approach (third edition) - Prentice Hall

Richard S. Sutton, Andrew G. Barto - Reinforcement Learning, an Introduction (second edition) - MIT Press