Modeling II
Math 350, Spring 2004

In this course, we will primarily be doing Empirical versus analytic modeling- that is, we'll be doing lots of data analysis.

• Course Syllabus
• Textbook: Introduction to Empirical Modeling. Below are PDF and PostScript (PS) versions. The postscript copies are clearer, but only use if you have a postscript viewer (all the computers in the Mathlab have postscript viewers). If you're on a Windows machine, use the PDF versions.
1. Contents, Chapter 1, Chapter 2 (PDF)
2. Contents, Chapter 1, Chapter 2 (PS)
3. Chapter 3 (Learning) in PDF
4. Chapter 3 (Learning) in PS
5. Ending of Chapter 3 plus Chapter 4 (in PDF)
6. Ending of Chapter 3 plus Chapter 4 (in PS)
7. New Chapter 4 and Chapter 5 (in PDF)
8. End of Chapter 5 through Chapter 7 (Only in PS)
This is a big file (about 4 MB) due to the eigenface images. You'll probably only want to download this from the Math Lab.
9. Neural Nets and RBFs (in compressed PS)
(Download and run "uncompress Ch11-12.ps.Z" to get the PS file).
10. Fourier Analysis (in compressed PS)
(Download and run "uncompress FourierChapter.ps.Z" to get the PS file).

Matlab

1. Letter E data
2. Card Game for Chapter 3 (right click and "Save Link Target As...")
3. The "banditE" program for reinforcement learning on the n-armed bandit problem
4. Driver for the banditE program (that is, use this script file to run banditE.m

Exam 1 Materials

1. Review Questions, Part I
2. Review Questions, Part II
3. Solutions to (12), Part II Review

Chapter 6 (KL) materials

• Homework and Descriptions
• M-file for first problem
• M-file for second problem
• Face data for third problem (NOTE: This is a practice data set containing 10 faces that are 64x64 so that the matrix Y is 4096 x 10).
• Movie data for last problem. This has 109 frames, each 120 x 160 pixels (so that Y1 is 19200 x 109). To view an image, you can use "imagesc(reshape(Y1(:,1),120,160))"

Chapter 11 (Radial Basis Functions) materials

• "edm.m": Matlab function to produce the Euclidean Distance Matrix
• Radial Basis Function Lab: Classification
• Script file for the Classification Problem
• Summary of RBFs
• Iris Data (For HW) (IRIS.NET)

Chapter 12 (Feed Forward Neural Networks

• A short script using Matlab to train a Neural Network (quickNNscript.m)
• Diabetes Data: Build a 8-15-15-2 network with tansigs, a training goal of 0.05 and use the algorithm "trainrp"

Final Project Materials: DUE by Tuesday, May 18th- No exceptions (you can turn them in sooner than that)

• Description of the Projects
• Comments and Matlab
• Data set 1: MenAndWomen.mat
• Data set 2: heart1.mat