Artificial Intelligence
Academic year 2017/2018
Course description
 Syllabus  Class schedule  Teaching material  Grading  Teaching assessment
 Useful links
 News  How to contact the instructor
Course description
This course introduces the students to some of the main
approaches, representations and techniques developed in the field of
Artificial Intelligence (AI) for designing systems capable to solve
complex information processing problems.
The following three areas are covered:
 search algorithms
 knowledgebased systems (knowledge representation and
inference, using
logical languages)
 machine learning (decision trees and artificial neural
networks)
The Lisp programming language is also presented.
Level
This is a graduate course under the M.Sc.
program in Electronic Engineering.
Prerequisites
Knowledge of elements of discrete mathematics (combinatorics),
computer architecture, and at least one programming language.
Academic period, duration and credits
 This course is taught during the first semester (October to
December)
 Class time (lectures and exercises): 50 h
 Credits (ECTS/CFU): 5
Syllabus
Introduction to AI
Historical notes, main AI approaches and subfields, AI achievements and
applications, philosophical and etchical issues.
Solving problems by searching
 Formulating
search problems: goal, states, actions, solutions, solution cost, state
space, search trees. Search algorithms. Properties of search
algorithms: optimality, completeness, computational complexity
 Uninformed search strategies: breadthfirst search,
depthfirst search, uniformcost search. Notes on depthlimited search,
Iterative deepening depthfirst search, bidirectional search
 Informed (heuristic) search strategies: bestfirst
search, greedy search,
A* search. Heuristic functions.
Knowledge representation and inference
 Introduction: motivating problems, knowledgebased systems
 Introduction to logic: argumentation (reasoning),
correctness of an argument, propositions, proofs, inference algorithms
 Logical languages, models
 Propositional logic: syntax, semantics, truth tables of
connectives, the model
checking inference algorithm. Inference rules and algorithms; the
forward and backward chaining inference algorithms
 Properties of inference algorithms: soundness,
completeness, computational complexity
 Predicate
(firstorder) logic: syntax and semantics; relationships between
quantifiers. Inference rules and algorithms; the forward and backward
chaining inference algorithms; notes on the resolution inference rule
and algorithm
 Application:
expert systems
Machine learning
 Hystorical notes and motivating problems
 Inductive learning (learning from examples)
 Supervised learning, automatic classification algorithms
 Decision trees: definition, learning algorithms
 Artificial neural networks. Hystorical notes. The
perceptron and its learning algorithm. Multilayer feedforward
networks; the backpropagation learning algorithm. Notes on deep neural
networks
 Notes
on the statistical formulation of supervised learning problems. Main
classifier evaluation methods: holdout, crossvalidation
The Lisp language
 Introduction: procedural (imperative) and declarative
(logical,
functional) programming paradigms and languages; hystorical notes on
Lisp
 Main builtin data types: atoms (numbers and symbols), lists
 Expressions and evaluation rules: selfevaluating atoms
(numbers,
special symbols), function calls and special expressions (lists)
 Main builtin functions: mathematics (=, +, , *,
/, sqrt, etc.), list processing (cons, first, rest; list, length,
nth, append, null), logic (equal, atom, listp, null)
 Main special expressions: quote; and,
or, not; cond (conditional expression)
 Userdefined functions: the defun special expression
 Binding values to symbols (variable assignments): the setq
and setf special
expressions
 Userdefined (structured) datatypes: the
defstruct special
expression
top
of the page
Class schedule
 Monday, 1719h, room BA
 Tuesday, 1718h, room BA
 Wednesday, 1113h, room BA
top of the page
Teaching material
Textbook
S.J. Russell, P. Norvig, Artificial
Intelligence: A Modern Approach, Pearson, 2010 (3rd ed.).
Several copies of the Italian edition are available at the Faculty
Library, and at other libraries of our University and outside
(S.J.
Russell, P. Norvig, Intelligenza Artificiale: un approccio moderno,
UTET, 1998; Pearson Education Italia, 2005).
Useful resources on artificial neural networks
 A classic textbook (several copies are available at the
Faculty
Library):
C.M. Bishop, Neural networks for
pattern recognition, Oxford University
Press, 1995
 ch. 1 (par. 1.1 to 1.6): Statistical
Pattern Recognition
 ch. 3: SingleLayer
Networks
 ch. 4: The
Multilayer Perceptron
 A
recent free online book which provides a very nice and gentle
introduction to artificial neural networks and deep neural networks:
M. Nielsen, Neural Networks and Deep Learning
Lecture slides and exercises
An extended
version of the slides projected during the lectures is
available.
Exercises:
Lisp resources
 Common Lisp implementations:
 Clozure
Common Lisp (CCL): open source, with a GUI for Mac OS only
 Lispworks
Personal Edition:
a free (limited) version of a commercial product
 CLisp:
open source, command line interpreter (not to be mistaken for the CLIPS expert system tool  see below)
 Jabberwocky: Javabased IDE (it requires a Lisp
interpreter, like CLisp)
 Common Lisp reference book: Guy L. Steele Jr.,
Common Lisp the Language, 2nd edition, Digital Press
Expert system tools
CLIPS is
a well known tool for building expert systems (not to
be mistaken for the CLisp
interpreter for the Lisp language  see above).
Chapters 16 of the User's
guide (cap. 16) are enough as a starting point. More details can
be found on the Basic Programming Guide.
Machine learning software
A
number of implementations of learning algorithms for decision trees and
artificial neural networks (beside many other classifiers) are
available. Among them:
 the Python scikitlearn
library (you may install before the package manager Miniconda,
which also installs Python)
 Weka: a Javabased software for designing
classifiers and carrying out experiments
 playground:
a useful tool for visualizing on a browser the inner working of
feedforward
multilayer artificial neural networks and of the backpropagation
learning algorithm, based on the open source TensorFlow
library
top of the page
Grading
The work for this course will consists of:
 one written examination containing questions and exercises
on all the topics on the syllabus
 one takehome project assignment involving the
implementation of one or more methods from one of the topics on the
syllabus (search algorithms, expert systems, machine learning, Lisp)
and their application to a given
problem; the projects can be done individually or in groups of two students
The exam will count for 10/30 of the grade, and the project
for 20/30.
top of the page
Teaching assessment
Teaching assessment questionnaires for the previous academic years are
available on the web site of the M.Sc. program in Electronic Engineering.
top of the page
Useful links
top of the page
NewsDecember 27, 2017
Results of the written test of Dec. 18.
December 18, 2017
Since Dec. 15 it is possible to fill out the teaching evaluation forms for all
the courses of the 1st semester, even by students who did not attend
the lectures. Filling out the form is mandatory
to enter for the exams. It is however recommended that all students
fill out the evaluation forms as soon as possible. The forms can be
accessed from your own Esse3
account.
November 30, 2017
From Jan. 23 to Feb. 6, 2018, prof. G.L. Marcialis will
give a
seminar on Biometric Technologies for Computer Security (24 hours, 2
ETCS credits if a final test is passed). More information and
registration instructions are available in the seminar web page.
top of the page
How to contact the instructor
Giorgio Fumera
Dept. of Electrical and Electronic Eng. (DIEE),
building B, 3rd floor
Phone: 070 675 5754
Email: surname AT diee DOT unica DOT it
top of the page
Most recent update: December 27, 2017
