APPLICATIONS OF 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
- 2014/2015
- Teacher
- FABRIZIO RIGUZZI
- Credits
- 6
- Didactic period
- Secondo Semestre
- SSD
- ING-INF/05
Training objectives
- The main goal of the course consists in enabling the students to design Artificial Intelligence systems (with an emphasis on knowledge-based systems) that solve problems such as automated planning, automated classification and machine learning, constraint satisfaction and optimization.
The main acquired knowledge refers to:
automated planning
inductive reasoning and machine learning techniques
constraint logic programming
The basic acquired abilities (that are the capacity of applying the acquired knowledge) will be:
ability of modeling planning problems and using automated planning systems to solve them
ability to analyze data for predicting purposes, including uncertain and relational data, extracting knowledge with machine learning techniques
ability to model constraint satisfaction and optimization problems and to solve them by applying solvers embedded in logic languages. Prerequisites
- The following concepts and knowledge provided by the course "Fundamentals of Artificial Intelligence" are mandatory:
first order logic
logic programming
problem solving by search
knowledge based systems architecture
knowledge representation models. Course programme
- The course consists of 48 hours of teaching partly in the classroom and partly in the laboratory.
Automated planning (8 hours) (Prof. Lamma) (planning as deduction, search in the state space and linear planners, search in the plane space and Partial Order Planning)
Inductive reasoning and machine learning techniques (24 hours) (Prof. Riguzzi) (probability theory recall, introduction to learning, concept learning and the general to specific ordering, decision trees, Bayesian networks, instance-based learning, propositional rule learning, first order rule learning, logical-probabilistic languages)
Constraint Logic Programming (16 hours) (Prof. Gavanelli) (Advantages and disadvantages of logic languages, constraint satisfaction and optimization problems, CLP(FD), consistency techniques, bound consistency, global constraints, sets in CLP(FD), search and optimization, applications, implementation of new solvers, CLP(R)). Didactic methods
- The course is composed of 48 hours of teaching partly in the classroom and partly in the computer laboratory.
The lectures will cover all the course topics and will include guided exercises on the computer.
Exercises in the laboratory will cover the use of Prolog to model and solve planning problems by resolution, the use of the Weka system to solve machine learning from examples problems, and the use of Constraint Logic Programming (CLP) systems such as ECLiPSe and SICStus Prolog to model and solve constraint problems. Learning assessment procedures
- The aim of the exam is to verify at which level the students achieved the learning objectives previously described.
The examination is composed of a written test and a computer programming test (available score 30/30).
The student can optionally sit an oral test on a specific topic (the oral test can add at most 2 points to the score of the written and laboratory tests).
Written test: questions and exercises on planning and machine learning (22 points)
Computer programming test: an exercise on constraint programming, requiring the authoring of a CLP program in the laboratory, in the same day of the written test, with which it composes the exam (8 points)
Oral test: discussion of a research article to be decided together with one of the instructors of the course. The assignment can be a four page summary of a research article or a small project. Before the discussion, the student must send the instructor a summary in English of the article of at most four pages (the test can add at most 2 points to the score of the written and laboratory tests).
The total score is the sum of points acquired in tne various tests. Reference texts
- T. M. Mitchell, “Machine Learning”, McGraw-Hill, 1997
Teachers’ handouts.
Texts for further reading:
S. J. Russel, P. Norvig: "Intelligenza Artificiale: Un approccio moderno", vol. 2, Pearson - Prentice Hall, 2005.
K. Apt, M. Wallace: Constraint Logic Programming using Eclipse. Cambridge University Press, 2007.
