CCE 580/CSI 505: Neural Networks Dr. Roger R.

Achkar (Beirut)

AMERICAN UNIVERSITY OF SCIENCE & TECHNOLOGY

FACULTY OF ENGINEERING
DEPARTMENT OF COMPUTER AND COMMUNICATIONS ENGINEERING

CCE 580/CSI 505– Neural Networks

Fall Term 2024-2025

INSTRUCTOR Dr. Roger Achkar

rachkar@aust.edu.lb

CLASS HOURS T 6:30 pm – 9:00 pm

Room: TBA

OFFICE HOURS T 5:00 pm - 6:15 pm

Otherwise by appointment
Room: Block A, Room 804

COURSE OBJECTIVES

The course will teach a variety of contemporary approaches to neural networks. It will
introduce the theory underlying these approaches, and how to implement them as computer
programs. The approaches to be covered will include such things as biological and statistical
foundations of neural networks, Perceptron, BP, SVM, RBFN and competitive learning.
Additionally, a brief introduction to information theory and its applications in neural
networks will be given.
In addition to that, MS students will be able to analyze published papers and to criticize them
which will let the future Ph.D. candidates discover the research world.

COURSE DESCRIPTION

The course will study artificial neural networks from the computer scientist's point of view.
Here, the new programming paradigms are - in short - learning by example, parallel
distributed processing and robust and fault tolerant (albeit not arbitrarily precise)
computation.
The course provides a thorough introduction to the backed methods of this new field: the
perceptron and linear separable functions, multilayer perceptrons, back-propagation, learning
algorithms for feedforward networks, variations and improvements of back-propagation,
generalization ability, recurrent networks: Hopfield Networks and Boltzmann Machines,
unsupervised learning, self-organizing.
Participants will learn to apply these methods and follow the current development in this
field.
Many advanced new topics will be presented through the analysis and presentation of recent
technical papers from relevant journals and conference proceedings.
The emphasis of the final project is to model a real life problem and build its neural network
system using any programming language.

-1-
CCE 580-Sec A: Neural Networks Dr. Roger Achkar (Beirut)

COURSE PREREQUISITES SENIOR STANDING

PREREQUISITES BY TOPICS

The student should have a good background of programming: CSI 250 – Programming II

COURSE CREDITS 3 Credit Hours

INSTRUCTION TECHNIQUE

Lecture will be used predominantly. These will be supported by handouts and problem sets.

REQUIRED TEXTS

Neural Networks: A Comprehensive Foundation by S. Haykin, Prentice Hall

REFERENCES
An Introduction to Neural Networks by K. Gurney, CRC Press

Elements of Artifical Neural Networks by K. Mehrotra et al., MIT Press

Principles of Neurocomputing for Science and Engineering by F. Ham, Mc Graw-hill

GRADE DISTRIBUTION

This course involves a number of activities, ranging from lectures, problem sets, quizzes, and
exams, to the midterm and final exams. All of these attributes of the course are intended to
help the student in developing his/her understanding of the material covered in CSI 505 and
in providing the department and the course instructor with information on how the student is
doing. Consequently, all of these activities are considered to be vital and will be taken into
consideration while assigning grades at the end of the term. When the student is being
assigned a course grade, it is imperative that this grade accurately reflects the student’s level
of achievement and his/her mastery of the material covered in CSI 505. A breakdown of the
weighting that will be used in making this assessment is as follows:

Class Attendance/ Participation ATT 05%

Homework Assignments /Drop Quizzes HMK 10%
Project PRJ 25%
Midterm Exam MID 25%
Final Exam FEX 35%

-2-
CCE 580-Sec A: Neural Networks Dr. Roger Achkar (Beirut)

Note: All the exams will be counted towards the final course grade based on the highest
grades attained. No make-up exams are allowed. Late assignments will be dealt with
according to the distributed course rules and regulations, which are governed by the
Department of Computer and Communications Engineering. Under severe conditions, only
the Final Exam is allowed to be petitioned; however, students have to realize that in the event
that a make-up Final Exam is granted, it would be unjust that the grantee becomes privileged
over his/her fellow colleagues in terms of extra-time to prepare for the exam and in
developing an idea about the contents of the exam. Accordingly, the make-up Final Exam will
carry an increased level of difficulty of at least 20% from the regular exam.

COURSE REQUIREMENTS

Students are required to read all assigned material prior to class and be prepared to discuss the
examples and provided assignments or handout by the professor. Discussion of these
examples and assignments will be the basis of the class participation grade.

ATTENDANCE

For legitimate reasons only, a student is allowed to absent him/herself for a maximum of 3
course lectures. However, any absence of more than three sessions will be counted toward
the 5% evaluation of the course grade. Absences beyond the specified maximum limit will
result in an automatic AW in the course and possibly other disciplinary measures. An
unexcused absence from a quiz will result in a zero on that quiz. Should the student be
absent from a lecture during which a problem set or a laboratory assignment is due, it is the
duty of the student to make certain that the homework assignment is handed in on time.
Failure to do so will incur a penalty on that assignment’s grade.
Note: 10 points will be deducted from the class attendance grade for every non excused
absence.

ACADEMIC INTEGRITY
The student should get familiar with the Students’ Handbook 2015-2017 of AUST. This is
distributed to the student at the beginning of every academic term. The student should also be
aware that plagiarism, abuse of laboratory facilities, and other sorts of academic dishonesty
are not tolerated and can result in unsympathetic penalties. The Department of Computer and
Communications Engineering (CCE) fully acknowledges the potential significance of
students studying together. In this sense, the CCE Department does not have any reservation
to this kind of collaboration, as long as all contestants are involved in all facets of the work,
and not with each individual contributing to a fraction of the assignment. Specifically, when a
student submits an assignment with his/her name on it, the CCE Department takes it for
granted that the details presented in the assignment are entirely the student’s own work, and
that this student has substantially participated in the creation of this work. If a portion of the
work has been conceived by collaborative work, that section should be highlighted and the
names of the students involved in this collaboration should be listed next to that section.
Note: No snacks or beverages, including water bottles, are allowed in the classroom.

PROBLEM SETS
For most part of the course, problem sets will be assigned on a weekly basis. Corrected
problem sets will be returned to the student at the beginning of the lecture that succeeds the
lecture of the homework assignment submission.

-3-
CCE 580-Sec A: Neural Networks Dr. Roger Achkar (Beirut)

COURSE OUTLINE BY TOPIC

The day-to-day topics to be covered in the table below may be adjusted as the subject proceeds, but
all examination dates are fixed, and problem set due dates are unlikely to change.

Week No.
 History and Structure of Neural Networks 1

 Neural Learning and Evaluation 2

 Supervised Learning : Single-Layer Networks 3

 Supervised Learning : Pocket Algorithm, 4

Adalines, Multiclass Discrimination

Midterm Exam: Tuesday, November 19, 2024 5

 Supervised Learning : Multilayer Networks I 6

 Supervised Learning : Backpropagation Algorithm, 7

 Accelerating the Learning Process

 Supervised Learning : Multilayer Networks II 9

 Supervised Learning : Prediction Networks, Radial Basis Functions, 10

 Polynomial Networks

 Unsupervised Learning : Winner-Take-All Networks, Vector Quantizers 11

 Unsupervised Learning : Counterpropagation Networks, 12

Adaptive Resonance Theory, Topologically Organized Networks

 Unsupervised Learning : Distance-Based Learning, 14

Neocognition, Principal Component Analysis Networks

 Associative Models : Hopfield Networks, Brain-Sate-in-a-Box Network 15

Associative Models : Boltzmann Machines, Hetero-associators

 Optimization Methods : using Hopfield Networks, 16

Simulated Annealing, Random Search. Evolutionary Algorithms T

PROJECTS Presentations 16 T 14/01/2025

READING PERIOD 16 Sa 18/01/2025

Su 19/01/2025

FINAL EXAM WEEK 17-18 M 20/01/2025

T 28/01/2025

DEVIATIONS: Minor deviations from the syllabus are a normal part of any adaptive teaching and learning
process
DISCLAIMER: Changes may be performed to the above syllabus without any prior notification.
Copyright © 2015-2025 American University of Science & Technology - Department of Computer and Communications Engineering.
All rights reserved.

-4-