ARTIFICIAL INTELLIGENCE
Academic year and teacher
If you can't find the course description that you're looking for in the above list,
please see the following instructions >>
- Versione italiana
- Academic year
- 2021/2022
- Teacher
- ALESSANDRO DRAGO
- Credits
- 6
- Didactic period
- Secondo Semestre
- SSD
- FIS/02
Training objectives
- The purpose of the course is to provide an introduction to artificial intelligence and machine learning. The techniques discussed in the course are now in common use in a variety of areas in pure and applied science. Implementations of neural networks and machine learning have been developed by huge companies such as Google and Facebook and are present in a variety of devices such as smartphones, navigators, instruments for medical diagnoses and radiation and particle detectors. The course aims to put these techniques inside the larger frame of the problems and of the techniques of Artificial Intelligence, also discussing the relation between these artificial structures and similar structures present in the brain. At the end of the course the student will acquire the ability to use some of the more relevant techniques of AI and the knowledge of the related problems.
Prerequisites
- The prerequisites are minimal, the mathematics involved is mainly linear algebra and simple calculus. No previous knowledge of the topics discussed in the course is assumed. It is useful if the students have at least some basic experience of programming, but it is not a mandatory prerequisite.
Course programme
- The course is divided into three sections.
Section 1 – Problem solving by uninformed and informed search (about 15 hours)
• Intelligent agents
• Uninformed search
• Euristic functions and informed search
Section 2 – Montecarlo techniques (about 12 hours)
• A short review of statistics
• Pseudo-random number generation
• Metropolis algorithm
• Travelling salesman problem
Section 3 – Neural networks (about 27 hours)
• Basic introduction to neuroscience
• The perceptron
• Multilayer perceptron
• Reinforcement learning Didactic methods
- The lectures will be divided between more theoretical lectures in which general ideas will be presented and more practical lessons in which algorithms implementing those ideas will be discussed and used. Computational tools such as Mathematica and Python will be used.
Learning assessment procedures
- The students will prepare an explicit implementation of one of the techniques studied during the course, showing that the student acquired the ability of using it. The oral exam will begin with discussing that implementation and will continue with other questions investigating some of the topics of general knowledge presented in the course.
Reference texts
- • Artificial Intelligence: an introduction. Russell-Norvig, Prentice Hall
• Numerical Recipes. Press et al.
• Neural Networks: an introduction. Mueller-Reinhardt-Strickland, Springer
• Theoretical Neuroscience. Dayan-Abbott, The MIT Press.
