Artificial Intelligence

A.Y. 2023/2024
Overall hours
Learning objectives
The course presents the theoretical foundations, the methodologies and the technologies of artificial intelligence for information and knowledge processing, with specific reference to neural networks, fuzzy systems, and evolutionary computing.
Expected learning outcomes
Understanding of the theoretical foundation, the methodologies, and the technologies of artificial intelligence, with specific reference to neural networks, fuzzy systems, and evolutionary computing.
Single course

This course can be attended as a single course.

Course syllabus and organization

Single session

Lesson period
First semester
Course syllabus
1. Introduction to Artificial Intelligence:
Technologies, nature-based methodologies, intelligent systems.

2. Neural Networks:
Biological background, Threshold Logic Units, definition, geometric interpretation, networks of Threshold Logic Units, training.
General neural networks, structure of neural networks, operation of neural networks, training neural networks.
Multilayer perceptrons, definition, function approximation, regression, gradient descent, error backpropagation, variants of gradient descent, Manhattan training, lifting, momentum term, self-adaptive error backpropagation, resilient error backpropagation, quick-propagation, weight decay, number of hidden neurons, deep learning, sensitivity analysis.
Radial basis function networks, definition, function approximation, initializing the parameters, training the parameters, generalized form.
Self-organizing maps, definition, learning vector quantization, neighborhood of the output neurons.
Hopfield networks, definition, convergence of the computations, associative memory, solving optimization problems, simulated annealing, Boltzmann machines.
Recurrent networks, representing differential equations, vectorial neural networks, error backpropagation in time.

3. Fuzzy Systems:
Introduction to fuzzy sets and fuzzy logic, natural languages and formal models, fuzzy sets, interpretation of fuzzy sets, fuzzy sets for modeling similarity, fuzzy sets for modeling preference, fuzzy sets for modeling possibility.
Representation of fuzzy sets, definition based on functions.
Fuzzy logic, propositions and truth values, t-norms and t-conorms, aggregation functions, basic assumptions, operations on fuzzy sets, intersection, union, complement, linguistic modifiers.
Extension principle, mappings of fuzzy sets, cartesian product and cylindrical extension.
Fuzzy relations, crisp relations, relations and deduction, chains of deductions, simple fuzzy relations, composition of fuzzy relations, fuzzy relational equations.
Similarity relations, similarity, fuzzy sets and extensional hulls, scaling concepts, fuzzy sets and similarity relations.
Fuzzy control, Mamdani controllers, Takagi-Sugeno-Kang controllers, fuzzy controller design, defuzzification methods, construction of a controller, logic-based controllers, control based on fuzzy relational equations, neuro-fuzzy control.
Fuzzy data analysis, fuzzy methods in data analysis, fuzzy clustering, c-means clustering, fuzzification by membership transformation, fuzzification by membership regularization, analysis of imprecise data using random sets, possibility theory and generalized measures, fuzzy random variables.

4. Evolutionary Algorithms:
Motivations, metaheuristics, biological evolution, simulated evolution, optimization problems.
Basic notions and concepts, building blocks of an evolutionary algorithm, optimization techniques, gradient ascent or descent, hill climbing, simulated annealing, threshold accepting, great deluge algorithm.
Elements of evolutionary algorithms, encoding of solution candidates, epistasis, fitness and selection, fitness proportionate selection, the dominance problem, vanishing selective pressure, adapting the fitness function, variance problem, rank-based selection, tournament selection, elitism, niche techniques, genetic operators, mutation operators, crossover operators, multi-parent operators, recombination operators, interpolating and extrapolating recombination.
Fundamental evolutionary algorithms, genetic algorithms, schema theorem.
Evolution strategies, selection, global variance adaptation, local variance adaptation, covariances, recombination operators.
Genetic programming, initialization, genetic operators, introns.
Multi-criteria optimization, weighted combination of criteria, pareto-optimal solutions, finding pareto-frontiers with evolutionary algorithms, parallelization.
Particle swarm optimization, influence of the parameters, multi-objective particle swarm optimization.
Ant colony optimization.
Prerequisites for admission
Knowledge of basic concepts of computer science, computer programming, mathematics (discrete and continuous).
Teaching methods
Teaching Resources
R. Kruse, C. Borgelt, C. Braune, S. Mostaghim, M. Steinbrecher, Computational Intelligence: A Methodological Introduction, Springer, 2016
Slides and videorecorded lectures are available on the course website:
Assessment methods and Criteria
Written exam aimed at verifying the student's knowledge and understanding of the subject. The written exam consists of theory questions. The duration of the exam is 2:00h. The mark is expressed in thirtieths and the grading will consider the correctness, completeness, and clarity of the answers to the questions. The exam is not sufficient if one or more answers are not sufficient. The exam is closed book.
INF/01 - INFORMATICS - University credits: 6
Lessons: 48 hours
Professor: Genovese Angelo
Appointment via e-mail
Office 6002 (6 floor)