All assignments are due on last day of class on May 5th


ASSIGNMENT 1  


1. From your book, page 53, #1, 2, 5.


 ASSIGNMENT 2 

2.  Implement an expert system in EXSYS to decide investment risk. 
    Use the decision tree of figure A.5, page 289 from your book as 
    source of knowledge.


 ASSIGNMENT 3  

Your book, numbers 3,4,5 , page 95.
Hand in the prereq structure diagram on paper (problem #3) and
develop two rule based systems, one for exercise 4 and one for 5. Hand
them in, on a diskette.


 ASSIGNMENT 4 
 
1. From your book, exercise 9 (a), page 183.

2. Repeat Assignment 2, but using the confidence factors approach this
   time. Consider the following rule confidence factors going from left to
   right on the decision tree branches:

   0.6, 0.9, 0.8, 0.7, 0.9, 0.6


ASSIGNMENT 5 

(A) Build a back propagation based neural model to learn the following
boolean function:

X1      X2      X3    OUT
1       1       1      1
1       0       1      1
1       0       0      1
1       1       0      1
0       1       1      0
0       1       0      0
0       0       0      0
0       0       1      0

(Note: use 6 patterns for training, 2 for testing. Experiment to find best architecture,
training set. Report the training error and testing error of your best network.)


ASSIGNMENT 6  

You are to implement an intelligent system which determines the amount of left
or right turn of the automobile driving wheel (in degrees),
in order for the automobile to follow a contour.
The input fuzzy variables are speed of the automobile and
the difference between the distances to the contour between the front
of the automobile and the rear.
A run by a human driver yielded these sensor measurements:

speed(miles)   difference(cm)    wheel turn(degrees)
10             +5                    15
50             +5                    30
10             +20                   40
50             +20                   60
15             -5                   -15
50             -5                   -30
15             -20                  -43
50             -20                  -65

Decided to try the NN approach and FL approach.

(A) Build a model using a BP Neural Network. Experiment to find the best 
achitecture, training set.

(B) Use the JFS Fuzzy Logic System to implement a fuzzy controler for 
this problem. You are to design the fuzzy sets, their membership functions,  
and test accuracy of your system at the end.

(C) which model is more accurate, the NN or the FLS? why do you think this is?
________________________________________________________________


______________________________________________________________