NEW COURSE:

 

Introduction to Neural Networks

CIS 350.002 – Spring 2002

Lab Page

 

meeting days:

Monday at 8:40A-10:30 (LAB CC200)
Wednesday at 8:40A-9:30 TL1A
Friday at 8:40A-10:30 TL1A

 

instructor:

Slobodan Vucetic, 323 Wachman Hall, phone: 204-5773

Office Hours: Wednesday 2:00 pm - 3:00 pm, or by appointment

 

lab assistant:

Nilesh Ghubade, e-mail: nileshg@temple.edu

 

Objective:

 

Neural networks have seen an explosion of interest over the last decade, and are being successfully applied across an extraordinary range of problem domains, in areas as diverse as finance, medicine, engineering, geology and physics. Their success can be attributed to a few key factors: (1) Power - neural networks are very sophisticated modeling techniques, capable of modeling very complex functions; (2) Ease of use - the level of user knowledge needed to successfully apply neural networks is much lower than would be the case using more traditional statistical methods.

 

This new course is designed to familiarize students with this fascinating new area of computer science. It will present students with commonly used neural network architectures and the corresponding learning algorithms, with an emphasis on understanding basic principles instead on insisting on too many mathematical details. To illustrate the practical utility of neural networks, case studies of their most successful real-life applications will be provided throughout the course. Lab sessions will give students the opportunity to gain the hands-on experience in applying neural networks on practical problems. To this goal, several tutorials on using Matlab and its toolboxes will be given as a part of the course.

 

While the course is targeted to computer science undergraduates, it can be suitable for students from statistics, engineering, science, business and other disciplines interested in learning basics of neural networks.

 

Prerequisites:

Senior or junior standing. Elementary computer programming skills (in an arbitrary programming language). Basic knowledge of linear algebra, calculus and statistics. In all other respects the course will be self-contained.

 

Textbook:

There is no required book for the course. Papers and handouts relevant to presented topics will be distributed as needed. The following is the list of recommended texts (they will be available on-reserve in Paley library):

· C.M. Bishop, Neural Networks for Pattern Recognition, 1995.

· T.M. Mitchell, Machine Learning, 1997

· R.O. Duda, P.E. Hart, D.G. Stork, Pattern Classification, 2000.

· R. Pratap, Getting Started with MATLAB 5, A Quick Introduction for Scientists and Engineers, 1998.
 

Topics:

· Introduction

- Review of Linear Algebra, Calculus, and Statistics relevant for study of Neural Networks

- Matlab tutorial (including using Neural Networks and other specialized Matlab toolboxes)

· Biological Foundations of Neural Networks

· Standard Architectures and Algorithms: Perceptron, Feedforward Neural Networks, Backpropagation

· Alternative Approaches: Self-Organizing Maps, Support Vector Machines, Radial Basis Functions, Ensemble Predictors

· Practical Issues: Data Preparation and Preprocessing, Model Building, Model Validation

· Applications: Finance, E-Commerce, Medicine, Biology, Image and Text Recognition, Engineering
 

 

Grading

Lab Quizzes, Homeworks, Midterm, Projects

 

final grade (could be adjusted):

Letter Grade

Total Points

A

> 90

B

81 – 90

C

71 – 80

D

61 – 70

F

< 61

 

Academic Honesty:

Academic honesty is taken seriously. You must write up your own solutions and code. For homework problems or projects you are allowed to discuss the problems or assignments verbally with other class members, TA, or instructor. You must acknowledge the people with whom you discussed your work. Any other sources (e.g. Web, research papers, books) used for solutions and code must also be acknowledged.