EE 555: Fundamentals of Intelligent Systems

Professor Mohamed A. El-Sharkawi
University of Washington

HW1: (due on April 9)

1. Write a steepest descent program to find the minimum of the cost function


2. Repeat using Newton-Raphson (NR) method

3. Discuss the differences between steepest descent and NR methods.

HW2: (due on April 16)

Write a projected gradient program to minimize the following Rosenbrock function.



Subject to the following constraint


HW3: (due on April 23)


y1=sin(a1 t)

y2=cos(a2 t)


Z=Y+v where v is white noise


1. Generate measurements: assume a1=1 and a2=2 and v=random(0.2).

2. Use the value of Z in step 1 to compute the frequencies of the sin and cos functions using steepest descent method. You should get the same values of a1 and a2 in step 1.


HW4: (due on May 7)

Write a matlab program to implement the error backpropagartion technique. Test your program by training a NN on sin and cos functions.


 HW5: (due on May 14)

Write a GA program.  Test your program by searching for the global minimum of the Rosenbrock function.  Use at least 2 parameters in the function






HW6: (due May 21)

Write a PSO program for the problem in HW5.


HW7 (due May 30):

Write an obstacles avoidance program based on attraction and repulsion forces only.  Assume the space is 100X100 pixels and the obstacles occupy 500 pixels in at least 25 clusters distributed randomly.  Move your agent from the lower left corner of the space to the upper right corner. Create a matlab movie of your solution.