MET CS 767 Fuzzy, Expert, Genetic and Neural Systems

Last updated 6/8/07.  Most recent updates are often in red.

Catalog Description red if updated recently Home Page of Eric Braude red if updated recently
Contacting Eric Braude red if updated recently Learning Objectives red if updated recently
Course Materials and References

red if updated recently

 Policies in all of Eric Braude’s Classes red if updated recently
Due Dates for Homework and Project red if updated recently Topics and class dates red if updated recently
Evaluation of Students red if updated recently Warning concerning plagiarism red if updated recently
Forums, Past and Present red if updated recently

Catalog Description

Prerequisites: Programming in Java or C++; recommended that students will have taken the core required courses for the MS degree in computer science.

Course description: Theories and methods for automating the solution of problems with inexact specifications, input, processing models or output. (e.g. text checkers, user profiles, help desks, intelligent agents). Expert systems, fuzzy methods, neural nets and genetic algorithms are described and compared. Algorithms and a term project are implemented using shells, C++ or Java.

Learning Objectives

  • Understand the goals, capabilities and limitations of soft computing
  • Be familiar with Expert Systems, Neural Nets, Fuzzy systems, and Genetic Algorithms
  • Be able to select among these given an application

Materials

The instructor will provide copies of presentation material for all classes.

Textbook:

“Soft Computing and Intelligent Systems Design: Theory, Tools and Applications” by Fakhreddine O. Karray and Clarence W De Silva; ISBN-10: 0321116178; ISBN-13: 978-0321116178

Students will probably want to acquire resources particular to the area on which they intend to focus.

Students will choose one or — preferably — a combination of two of the four areas in which to design and execute a project, and can purchase recommended literature accordingly.

Past students in this course have begun to develop a list of references and tools at http://metcs.bu.edu/~ebraude/767/articles/index.htm .  See also the forums.  See also http://jooneworld.com/index.html for a good neural net framework.

Evaluation of Students

The course will consist of homework and a project, weighted as follows.

  • Homework:                   25%
  • Project:                         75%

The project will be in three phases, weighted as follows:

phase 1 (problem statement): — 1/6

phase 2 (analysis & design): — 1/3

phase 3 (implementation and critical review): –1/2

Parts of assignments are evaluated equally unless otherwise stated.

Students may be permitted to substitute parts of these with a special paper, approved in advance within the first 3 weeks. Late homework without a reason why it was impossible will not be accepted. If there is such impossibility, the work will be graded on a pass/fail basis.  Reasons should be clearly written on the front of the paper.  The fax (617) 353-2367 should be used if you cannot be at class.

Students are required to make a presentation on their project.  The suggested organization is as follows.

1.       Project goals (application and learning)

2.      Method and design

3.      Outcomes: challenges, difficulties and problems

4.      Outcomes: successes

5.      What would be required to make real

Syllabus

Since these are cutting-edge topics, the syllabus may be adjusted somewhat during the semester.

Class

Num

 Date Topic Notes and Related Reading Comments

The due dates mentioned below are only approximate.  For finalized due dates click here.
1 5/22 Introduction to Soft Computing I

Contrast expert systems and fuzzy systems

 Karray-DeSilva 1 think about project
2 5/29 Introduction to Soft Computing II

Contrast expert systems, neural nets, fuzzy systems, and genetic algorithms

 Karray-DeSilva 1 Project: Phase 1 assigned
3 6/5 Introduction to Expert Systems

Define expert systems; knowledge representation

Inference in Expert Systems: Part I

Using rules & decision trees
4 6/12 Inference in Expert Systems: Part II

Knowledge Acquisition; Processing uncertainty; preliminary comparison with fuzzy systems

 Phase 1 due;

phase 2 assigned
5 6/19 Introduction to Fuzzy Systems

Define; basic set theory; describe applications

 Karray-DeSilva 2
6 6/26 To be decided Phase 3 assigned

Phase 2 due
7 7/3 Implementation of Fuzzy Systems

Architectures & tools

 Karray-DeSilva 3
8 7/10 Introduction to Neural Nets

Basic architectures

 Karray-DeSilva 4
9 7/17
Backpropagation
Define and use the algorithm
 Karray-DeSilva 5 and 6
10 7/26 Introduction to Genetic Algorithms

Define and use genetic algorithms

 Karray-DeSilva 8
11 7/31 Genetic Algorithms and Evolutionary Computations Karray-DeSilva 8 Project: Phase 3 due
12 8/7 Student Presentations and Demonstrations

Warning Concerning Plagiarism

The College has serious penalties for plagiarism, including expulsion from the degree program. Please be very careful not to use the work of others without very clear and specific acknowledgement.
e-mail, see or call me if you have any doubts. In any case, clearly acknowledge all sources in the context they are used, including code, of course.

Forum

Past forums: 1999 ,2000, Summer 2002, Spring 2003, Fall 2003, Summer 2005

Summer 2007:

Group name:    767Su07

Group home page:    http://groups.yahoo.com/group/767Su07

Group email:    767Su07@yahoogroups.com