2. Professors
  3. PALOMBA Fabio
  4. Teaching

FUNDAMENTALS OF ARTIFICIAL INTELLIGENCE

Fabio PALOMBA FUNDAMENTALS OF ARTIFICIAL INTELLIGENCE

0512100055
COMPUTER SCIENCE
EQF6
COMPUTER SCIENCE
2022/2023

YEAR OF COURSE 3
YEAR OF DIDACTIC SYSTEM 2017
AUTUMN SEMESTER
CFUHOURSACTIVITY
648LESSONS
FABIO PALOMBA T Curriculum
Objectives
THE COURSE IS MEANT TO INTRODUCE PROBLEM SOLVING PRINCIPLES AND TECHNIQUES ADOPTED IN THE FIELD OF ARTIFICIAL INTELLIGENCE (WITH SPECIAL FOCUS ON KNOWLEDGE BASED SYSTEMS AND LOGIC BASED METHODS).

KNOWLEDGE AND UNDERSTANDING
STUDENTS ARE EXPECTED TO GAIN KNOWLEDGE AND GOOD AWARENESS OF PROBLEM SOLVING CONCEPTS AND METHODS IN ARTIFICIAL INTELLIGENCE.

ABILITY TO APPLY KNOWLEDGE AND UNDERSTANDING
STUDENT IS EXPECTED TO BE ABLE TO DEFINE AND DEVELOP INFERENCE ENGINES THROUGH IMPERATIVE OR DECLARATIVE PROGRAMMING LANGUAGES

THE STUDENT WILL BE ABLE TO DEFINE AND IMPLEMENT SOLVERS FOR KNOWLEDGE-BASED SYSTEMS THROUGH IMPERATIVE OR DECLARATIVE LANGUAGES:
- ABILITY TO RECOGNIZE PROBLEMS SOLVABLE THROUGH THE ADOPTION OF ARTIFICIAL INTELLIGENCE ALGORITHMS;
- ABILITY TO IDENTIFY WHICH IS THE MOST SUITABLE SOLUTION TO THE RESOLUTION OF AN ARTIFICIAL INTELLIGENCE PROBLEM AMONG THE VARIOUS POSSIBLE ALTERNATIVES.
- ABILITY TO MODEL AND SOLVE AN ARTIFICIAL INTELLIGENCE PROBLEM;
- ABILITY TO IMPLEMENT A SOLUTION TO AN ARTIFICIAL INTELLIGENCE PROBLEM THROUGH THE USE OF METHODOLOGIES AND TOOLS AVAILABLE ON THE MARKET.
Prerequisites
THE STUDENT MUST HAVE ACQUIRED BASIC KNOWLEDGE AND SKILLS ON MATHEMATICAL LOGIC, PROBABILITY, AND PROGRAMMING.
Contents
THE STUDENT WILL ACQUIRE KNOWLEDGE AND SKILLS ON THE MAIN CONCEPTS AND METHODS REPRESENTING THE GROUND FOR THE RESOLUTION OF ARTIFICIAL INTELLIGENCE PROBLEMS.

SPECIFICALLY:

- MODELING AND REPRESENTATION OF ARTIFICIAL INTELLIGENCE PROBLEMS;
- SMART AGENTS;
- SEARCH-BASED ALGORITHMS;
- ADVERSARIAL SEARCH-BASED ALGORITHMS;
- HEURISTIC AND META-HEURISTIC ALGORITHMS;
- SUPERVISED LEARNING ALGORITHMS;
- UNSUPERVISED LEARNING ALGORITHMS;
- METHODS OF ENGINEERING AND EVALUATION OF ARTIFICIAL INTELLIGENCE ALGORITHMS;
- USABILITY OF ARTIFICIAL INTELLIGENCE ALGORITHMS;

THE COURSE WILL PROVIDE BASIC KNOWLEDGE ON THE FOLLOWING TOPICS:

KNOWLEDGE REPRESENTATION (6 HOURS):
- BACKGROUND ON COGNITIVE PSYCHOLOGY, ON THE BORN AND RISE OF ARTIFICIAL INTELLIGENCE, AND ON THE RELATION BETWEEN COGNITIVE PSYCHOLOGY AND ARTIFICIAL INTELLIGENCE;
- SMART AGENTS AND KNOWLEDGE REPRESENTATION STARTING FROM A TEXTUAL DESCRIPTION.

RESOLUTION OF PROBLEMS THROUGH SEARCH-BASED ALGORITHMS (16 HOURS):
- INFORMED SEARCH TECHNIQUES (BREADTH SEARCH, UNIFORM-COST SEARCH, DEPTH SEARCH, LIMITED DEPTH SEARCH, DEPTH-FIRST SEARCH, BIDIRECTIONAL SEARCH);
- UNINFORMED SEARCH TECHNIQUES (BEST-FIRST GREEDY SEARCH, A* SEARCH, BEAM SEARCH, ITERATIVE DEEPENING A*, SIMPLIFIED MEMORY BOUNDED A*);
- LOCAL SEARCH TECHNIQUES (HILL-CLIMBING ALGORITHM, SIMULATED ANNEALING ALGORITHM, LOCAL BEAM SEARCH, GENETIC ALGORITHMS);
- ADVERSARIAL SEARCH TECHNIQUES (NASH EQUILIBRIUM, PARETO OPTIMAL, MINIMAX ALGORITHM, ALFA-BETA PRUNING, ADAPTATION OF GAME THEORY ALGORITHMS TO GAMES WITH MORE PLAYERS).

RESOLUTION OF PROBLEMS THROUGH LEARNING ALGORITHMS (20 HOURS):
- LEARNING THEORY (SUPERVISED, UNSUPERVISED, AND CLUSTERING ALGORITHMS);
- DECISION AND ERROR THEORY (ERROR, BIAS, VARIANCE);
- LIFECYCLE MODELS FOR ARTIFICIAL INTELLIGENCE ALGORITHMS;
- EXTRACTION, PRE-ELABORATION, AND QUALITY OF DATA (DATA CLEANING AND PRE-PROCESSING ALGORITHMS);
- EXTRACTION OF FEATURES FROM DATA (FEATURE EXTRACTION AND SELECTION TECHNIQUES);
- CLASSIFICATION ALGORITHMS (DECISION TREES, BAYESIAN NETWORKS);
- REGRESSION ALGORITHMS (SINGLE LINEAR REGRESSION, MULTIPLE LINEAR REGRESSION);
- CLUSTERING ALGORITHMS (K-MEANS, VARIANTS OF K-MEANS, DBSCAN);
- PERFORMANCE ANALYSIS (CONFUSION MATRICES, METRICS TO EVALUATE MACHINE LEARNING SYSTEMS).

USABILITY OF ARTIFICIAL INTELLIGENCE ALGORITHMS (6 HOURS):
- BASIC NOTIONS ON THE PROVISION OF INFORMATION COMING FROM ARTIFICIAL INTELLIGENCE ALGORITHMS;
- DESIGN AND DEVELOPMENT OF CONVERSATIONAL AGENTS.

ADDITIONAL LAB SESSIONS THAT WILL BE INCLUDED IN THE COURSE:
- INTRODUCTION TO THE PYTHON PROGRAMMING LANGUAGE AND FRAMEWORK TO SOLVE ARTIFICIAL INTELLIGENCE PROBLEMS;
- INTRODUCTION TO THE WEKA FRAMEWORK TO SOLVE ARTIFICIAL INTELLIGENCE PROBLEMS;
- PRESENTATION AND DISCUSSION OF CASE STUDIES.
Teaching Methods
THE COURSE INCLUDES 48 HOURS OF FRONTAL LECTURES AND EXERCISES (6 ECTS), WITH THE GOAL OF PRESENTING THE ENVISIONED CONCEPTS AND DEVELOPING THE CAPABILITIES NEEDED FOR THE RESOLUTION OF ARTIFICIAL INTELLIGENCE PROBLEMS THROUGH THE USE OF THE (SEMI-)AUTOMATIC TOOLS DISCUSSED IN THE CONTEXT OF THE COURSE.
Verification of learning
THE VERIFICATION OF THE SKILLS ACQUIRED BY THE STUDENT WILL BE VERIFIED THROUGH AN EXAM, WITH EVALUATION IN THIRTIETHS. THE EXAM INCLUDES A WRITTEN EXAM AS WELL AS THE DEVELOPMENT OF A PROJECT.

- THE WRITTEN EXAM HAS THE GOAL OF VERIFYING THE THEORETICAL SKILLS ACQUIRED BY THE STUDENT ON THE USAGE OF ARTIFICIAL INTELLIGENCE METHODOLOGIES AND TECHNIQUES;

- THE PROJECT HAS THE GOAL OF EVALUATING THE COMPLETENESS AND CORRECTNESS OF A PROJECT RELATED TO THE APPLICATION OF ARTIFICIAL INTELLIGENCE METHODOLOGIES AND TECHNIQUES IN REAL CONTEXTS. FURTHERMORE, IT HAS THE GOAL OF ASSESSING THE LANGUAGE SKILLS AS WELL AS THE ABILITY TO PROPERLY MOTIVATE THE CHOICES DONE DURING THE PROJECT DEVELOPMENT. AT THE END OF THE PROJECT, THE STUDENT WILL DELIVER (1) A REPORT CONTAINING THE PROJECT DOCUMENTATION - DEVELOPED USING LATEX - AND (2) A 15-MINUTE PRESENTATION - DEVELOPED USING KEYNOTE, POWERPOINT, OR GOOGLE PRESENTATION).

THE FINAL EVALUATION WILL TAKE THE OUTCOME OF THE TWO PARTS INTO ACCOUNT.
Texts
AS FOR THE SEARCH ALGORITHMS PART, THE RECOMMENDED BOOK IS:

- S. J. RUSSELL, P. NORVIG. “ARTIFICIAL INTELLIGENCE: A MODERN APPROACH”, PRENTICE HALL, 2002.

AS FOR THE LEARNING ALGORITHMS PART, THE RECOMMENDED BOOKS ARE:

- S. J. RUSSELL, P. NORVIG. “INTELLIGENZA ARTIFICIALE: UN APPROCCIO MODERNO”, VOLUME 1 (TERZA EDIZIONE, 2010) E VOLUME 2 (SECONDA EDIZIONE, 2005), PEARSON EDUCATION ITALIA.

- A. GERON. "HANDS-ON MACHINE LEARNING WITH SKIKIT-LEARN, KERAS & TENSORFLOW", 2ND EDITION, O'REALLY.

- A. BURKOV, “MACHINE LEARNING ENGINEERING”, PAPERBACK.

FURTHER RECOMMENDED READINGS:

1. C. M. BISHOP. “PATTERN RECOGNITION AND MACHINE LEARNING”, SPRINGER SCIENCE, NEW YORK, 2006.

2. DUDA, R. O., HART, P. E., & STORK, D. G. (2012). PATTERN CLASSIFICATION. JOHN WILEY & SONS.
More Information
ATTENDING THE COURSE IS NOT MANDATORY BUT STRONGLY RECOMMENDED. STUDENTS MUST BE READY TO ATTEND THE COURSE ACTIVELY, THROUGH THE INTERACTION WITH THE LECTURERS AS WELL AS THE INDIVIDUAL STUDY OF THE MATERIAL TAUGHT DURING THE LECTURES. A SATISFACTORY PREPARATION WHICH LEADS TO PASSING THE EXAM WILL CONSIST OF AN AVERAGE INDIVIDUAL STUDY OF TWO HOURS FOR EACH HOUR OF LECTURE AND AN AVERAGE OF ONE HOUR DEVOTED TO THE DEVELOPMENT OF THE PROJECT. BY DESIGN, THE COURSE EXPECTS A STRONG PREDISPOSITION TO LEARNING NEW SOFTWARE INSTRUMENTS FOR THE DEVELOPMENT OF ARTIFICIAL INTELLIGENCE MODULES.

THE DIDACTIC MATERIAL WILL BE MADE AVAILABLE ON THE E-LEARNING PLATFORM OF THE DEPARTMENT.
  BETA VERSION Data source ESSE3 [Ultima Sincronizzazione: 2024-08-21]