EE 4745: Neural Computing

Summer 2002

Last updated: July 29, 2002 (2:30 PM)

Professor:         J. Ramanujam (Ram)
Office:                 345 EE
Phone:                578-5628
E-mail:                jxr AT ece DOT lsu DOT edu
Office hours (till July 26) :    Mon, Wed    9:45 -11:45;      Thu    9:45 - 10:45
Exam Week Office hours:    Mon (July 29)     9:45 -11:45;      Tue (July 30)     8:45 - 11:00
Final Exam (Comprehensive): :    Wednesday July 31, 1:30 PM - 3:30 PM Room 145 EE Building

Course time and place: 8:40-9:40 MTWTF in Room EE 145 (usually) or EE 117

Links:

Announcements:
- Extra Problem Set (not for grade) for Sections 14 and 17 now available.
- In Program 3, you need to do two sets of experiments. In the first set of experiments, use tansig for the hidden layer and purelin for the output layer as the transfer functions. In the second set of experiments, use logsig for the hidden layer and purelin for the output layer as the transfer functions. See Section 5 of the Neural Networks Toolbox Manual ( PDF file ) ( html ) for using backpropagation.
- The computer lab in the ECE building will be open Sunday, July 28th 2002 from noon to 4 PM.
- In addition, several of the following LSU public access labs that have MATLAB (not sure about the neural networks toolbox though; chances are they have the neural networks toolbox which is very popular) are open both Saturday and Sunday. Check the Office of Computing Services webpage for hours.
```
    ---------------------------------------
         Location                Semester 
    ---------------------------------------
    CEBA Reading Room           Summer 2002 
    Horticulture 225            Summer 2002 
    Union 332                   Summer 2002 
    CEBA 1302                   Summer 2002 
    Coates 263                  Summer 2002 
    Middleton 101               Summer 2002 
    Middleton 141               Summer 2002 
    Middleton 241               Summer 2002 
    Williams 3rd floor          Summer 2002 
    ---------------------------------------
```
- Homework 5 is now due on July 29, 02.
- Program 3 due on July 29, 02 is available.
- Here is a list of questions and clarifications concerning Program 2. See new addition on plotting.
- Program 2 due on July 29, 02 is available.
- Homework 5 due on July 29, 02 is available.
  Problem 8: Change Line 2 from "Page 13-9" to "Page 13-12".
  Problem 10: Change all references in this question to "Page 13-9" to read "Page 13-17".
Homework
- Extra Problem Set (not for grade) for Sections 14 and 17 now available.
- Homework 5 is now due on July 29, 02.
- Homework 5 due on July 29, 02 is available.
  Problem 8: Change Line 2 from "Page 13-9" to "Page 13-12".
  Problem 10: Change all references in this question to "Page 13-9" to read "Page 13-17".
- Homework 4 due on July 15, 02 Scanned copy of Homework 4 available (one file per page):
  Page 1, Page 2, Page 3, Page 4, Page 5.
- Clarification on HW 4:
  - For Question 1: in part (c) testing is done using vector p not its normalized form for both the Hebb rule and the Pseudoinverse rule. See comment below concerning what is done with Hebb rule.
  - For Question 2: In using the Hebb rule, the standard thing to do is to use the normalized forms of vectors both in training and in testing. I just wanted to make this clear. That is why I had written "(use normalized form in testing)" at the end of part (b)
- Available: Solution for Homework 3.
- Homework 3 due on July 11, 02
- Homework 2 due on July 1, 02
- Homework 1 due on June 25, 02
Programs
- Program 3 due on July 29, 02 is available.
- Here is a list of questions and clarifications concerning Program 2.
- Program 2 due on July 29, 02 is available.
- Program Set 1 due on July 11, 02
Syllabus
Lecture Overheads (Sets 1-14, 17 and 18 available as of July 5th)

Reference texts:

On-line books:
M. Hagan, H. Demuth and M. Beale (1996), Neural Network Design, PWS Publishing, Boston, MA.
MATLAB (version 6.1) and MATLAB Neural Networks Toolbox (version 4.0):
C. Looney (1997), Pattern Recognition using Neural Networks, Oxford University Press, New York, NY.
J. Principe, N. Euliano, W. Lefebvre (2000), Neural and Adaptive Systems, John Wiley & Sons, New York, NY.
C. M. Bishop (1995), Neural Networks for Pattern Recognition, Oxford University Press, Oxford, UK.
J. Hertz, A. Krogh, R. Palmer (1991). Introduction to the Theory of Neural Computation, Addison-Wesley, Redwood City, CA.
L. Fausett (1994), Fundamentals of Neural Networks: Architectures, Algorithms, and Applications, Prentice-Hall, Englewood Cliffs, NJ.