Module Overview

Foundations of Intelligent Systems

This course provides the introduction to the concepts of logic, graph search and knowledge representation paradigms necessary to provide an understanding of more complex intelligent systems.

The aims of this module are to:

Give the student an understanding and appreciation of the various formalisms and concepts underlying intelligent systems.

Enable the student to apply those formalisms in the context of more complex systems.

Allow the student to understand the logic programming paradigm and be able to state solutions to simple problems using this paradigm.

Module Code

INFS9404

ECTS Credits

5

*Curricular information is subject to change

Propositional and first-order predicate logic: syntax and semantics, inference rules, unification and pattern-matching.
Problem-solving as a search procedure: state space search, fundamentals of graph theory, search strategies: forward- and backward-chaining, backtracking.
Uninformed graph-search algorithms: depth-first, breadth-first.
Heuristic graph-search algorithms: hill-climbing, best-first search, determining suitable heuristics, horizon effect.
Game-playing or competitive graph-search algorithms: MINIMAX, Alpha-Beta pruning.
Higher-level search techniques: recursive search, pattern-directed search.
Knowledge representation techniques: logical, procedural, network, structured.
Production system model: components, functions.
Models of reasoning: rule-based, model-based, case-based.
Expert systems: development process, roles of participants, components, problems amenable to expert system solution, use of expert system shells.
Uncertainty: necessity for mechanisms to deal with uncertainty, confidence measures, statistical methods, belief measures, fuzzy logic, nonmonotonic logic.
Constraint-based reasoning, constraint programming.
Overview of logic programming paradigm, and syntax and implementation of a logic programming language.
Planning: means-ends analysis, linear and non-linear planning algorithms.
Automated reasoning and theorem proving.
Intelligent agents: classes of agents, characteristics of environments, multi-agent systems, applications.
Stochastic formalisms and their applications in learning: Finite-state machines, Markov and Hidden Markov Models, dynamic Bayesian networks.
Reinforcement learning: concepts of rewards, policies.
Swarm intelligence: concepts, ant colony optimisation, particle swarm optimization, applications.

Propositional and first-order predicate logic:

Syntax and semantics, inference rules, unification and pattern-matching.Automated reasoning and theorem proving, resolution.

Problem-solving as a search procedure:

State space search, fundamentals of graph theory, search strategies: forward- and backward-chaining, backtracking.

Graph search algorithms:

Uninformed graph-search algorithms: depth-first, breadth-first.Heuristic graph-search algorithms: hill-climbing, best-first search, determining suitable heuristics, horizon effect.Game-playing or competitive graph-search algorithms: MINIMAX, Alpha-Beta pruning.Higher-level search techniques: recursive search, pattern-directed search.

Overview of knowledge representation techniques:

Logical, procedural, network, structured.

Structured reasoning models:

Production system model: components, functions.Models of reasoning: rule-based, model-based, case-based.Rule-based (expert) systems: development process, roles of participants, components, problems amenable to expert system solution, use of expert system shells.

Reasoning under uncertainty:

Necessity for mechanisms to deal with uncertainty, confidence measures, statistical methods, belief measures, fuzzy logic, nonmonotonic logic.

Constraint-based reasoning:

Constraint-based reasoning, constraint programming.

Planning:

Means-ends analysis, linear and non-linear planning algorithms.Planning under uncertainty, notions of rewards,optimal policies, discounting, Bellman equations.

Agent-based systems:

Intelligent agents: classes of agents, characteristics of environments, multi-agent systems, applications.Swarm intelligence: concepts, ant colony optimisation, particle swarm optimization, applications.

Stochastic formalisms and their applications in learning:

Finite-state machines, Markov and Hidden Markov Models, dynamic Bayesian networks.Reinforcement learning: concepts of rewards, policies.

Logic programming:

Overview of logic programming paradigm, and syntax and implementation of a logic programming language.

The learning methods used for this module will be a combination of lectures and practical problem-solving exercises, with a final written assessment under exam conditions. The practical assessments will ask students to implement simple search and knowledge representation problems using languages based on logic programming paradigms.

Module Content & Assessment
Assessment Breakdown %
Formal Examination80
Other Assessment(s)20