Descent, Newton's Method and Method Comparison From Chapter 9 and The Marquardt Demo

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 1

AS-74.

3115 Neuro-Fuzzy Computing in Automation


Spring 2010
Exercise 5
1a) Run the Matlab Neural Networks Generalization demo. What kind of phenomena typical for
neural network training can be discovered? Type nnd to open the Neural Network Design
demonstrations and select Generalization from the Chapter 11.
1b) Compare the different training algorithms by running the demonstrations Steepest
Descent, Newtons method and Method comparison from Chapter 9 and the Marquardt demo
from Chapter 12.

2) A typical sequence of fading signal in a mobile communications system is depicted in the


figure below. Design a neural network-based predictor to produce one-step-ahead prediction of
the fading signal. Use the given two sequences from the given signal as training and validation
data sets. The data is stored in file fading.mat which can be downloaded from the course
webpage.
10

Fading Power Level

-5

-10

-15

-20

-25

-30

-35

-40

500

1000

1500

2000
2500
3000
Time in Samples

3500

4000

4500

5000

3) Design an instrument that determines the serum cholesterol levels from the spectral
measurements of a blood sample. We have a total of 264 patients for which we have
measurements of 21 wavelengths of the spectrum. For the same patients we also have
measurements of HDL (high-density lipoprotein), LDL (low-density lipoprotein), and VLDL
(very low-density lipoprotein) cholesterol levels, based on serum separation. Design a neural
network for the task and use Principle Component Analysis for the preprocessing of inputs. You
can load the data by typing load choles_all.

You might also like