OVERVIEW

Soft Computing aims to introduce a set of Artificial Intelligence computational techniques which are based on the emulation of biological processes.

In particular, we will introduce neural networks mimicking human brain skills to learn and generalize, evolutionary computation mimicking adaptation of biological species to their own environment, fuzzy clustering  mimicking human ability to handle vague concepts, and swarm intelligence based on the model of intelligent cooperative behavior of some animals. For each topic covered, a laboratory exercise using MATLAB is planned.

Both the theoretical lectures and the lab sessions will be held in person.

AIMS AND CONTENT

LEARNING OUTCOMES

The course aims to describe the main numerical computation techniques based on the emulation of biological processes. The goal is to provide students with modern computational tools within the Artificial Intelligence domain such as neural networks with and without memory, self-organizing maps, genetic algorithms, evolutionary computation, and swarm intelligence.

AIMS AND LEARNING OUTCOMES

The overall objective of the module is to provide students with modern computational techniques, based on the emulation of successfull biological processes, for the solution of classification, clustering, optimization and forecasting problems. In particular, specific objectives are aimed at the acquisition of knowledge and skills in the field of:

• Supervised and unsupervised learning techniques for classification, clustering and prediction problems (neural networks);
• Clustering techniques (fuzzy clustering);
• Global optimization techniques (evolutionary computing, swarm intelligence)

Specifically, upon completion of the course, students will know how to:

• Design and implement the architecture of single-layer and multilayer neural networks;
• Implement genetic algorithms and compare their performance under varying selection, reproduction and mutation strategies adopted;
• Solve unsupervised clustering problems following fuzzy approaches and infer a posteriori information characterizing the identified classes;
• Implement the swarm intelligence algorithm ACO (Ant Colony Optimization).

PREREQUISITES

All slides used during the lectures, as well as other teaching materials, will be available on Aulaweb. In general, the notes taken during the lectures and the materials provided on Aulaweb are sufficient for exam preparation.

TEACHING METHODS

The course includes theoretical lectures, totaling 26 hours, and three laboratory sessions, totaling 26 hours. Both the lectures and the lab sessions will take place in person. Attendance at the laboratory sessions is highly recommended. At the beginning of each lab activity, a brief theoretical introduction will be given to review the computational techniques to be used and outline the steps to follow. During the practical part, students—working in groups of two or three and supported by the instructors—will be required to implement the described techniques. At the end of each lab session, students must submit their code along with a brief report detailing the methods used and the results obtained. The organization and schedule of the individual lab sessions will be communicated directly by the instructors during the lectures.

SYLLABUS/CONTENT

Course Topics

Neural networks

• General overview
• Single- and Multi-layer perceptron
• Introduction to Recurrent Neural Networks and Self-Organizing Maps
• Convolutional Neural Networks (CNNs)

Evolutionary computation

• General overiview
• Genetic algorithms (alphabets for encoding individuals, fitness, selection criteria for reproduction, crossover and mutation operators, natural selection)

Swarm Intelligence

• Particle Swarm Optimization
• Ant Colony Optimization

RECOMMENDED READING/BIBLIOGRAPHY

All slides used during the lectures, as well as other teaching materials, will be available on Aulaweb. In general, the notes taken during the lectures and the materials provided on Aulaweb are sufficient for exam preparation.

TEACHERS AND EXAM BOARD

LESSONS

LESSONS START

In accordance with the academic calendar approved  by the Consiglio di Corso di Studi.

Class schedule

The timetable for this course is available here: Portale EasyAcademy

EXAMS

EXAM DESCRIPTION

During the semester, three laboratory sessions will be held, each with a set deadline for the submission of the corresponding code, including a brief report on the methodologies used and the results obtained. A positive evaluation of the lab work is a prerequisite for admission to the final oral exam. For students who have attended at least 75% of the lab hours, the oral exam will focus on the theoretical topics covered in the course. For all others, it will also include more technical aspects of the lab exercises.

Students with disabilities or specific learning disorders (DSA) are reminded that, in order to request accommodations for exams, they must follow the instructions detailed on the following page: https://unige.it/disabilita-dsa/studenti-disturbi-specifici-apprendimento-dsa. In particular, requests for accommodations must be made well in advance (at least 7 days before the exam date) by contacting the Professor and copying both the School's designated contact person and the relevant office (see instructions).

ASSESSMENT METHODS

The laboratory assessments are aimed at testing the practical skills acquired for the solution of the posed problems. They will be evaluated on the basis of the following criteria:

• accuracy and optimization of the code
• accuracy and presentation of the results (images, graphs, tables ...)
• comments on the procedures followed and on the results obtained

The oral exam is finally aimed at assessing the ability to communicate the knowledge acquired in a clear and competent manner

FURTHER INFORMATION

Students with DSA certification (specific learning disabilities), disability or other special educational needs are advised to contact the teacher at the beginning of the course to agree on teaching and examination methods that, in compliance with the teaching objectives, take account of individual learning arrangements and provide appropriate compensatory tools.