Neural Networks: Notes and Matlab Code

• Matlab function banditE.m that implements the n-armed bandit problem using epsilon-greedy algorithm.
• Script to run banditE.m
• Matlab function banditS.m that implements the softmax strategy
• Script to run banditS.m
• Matlab function banditP.m that implements the pursuit strategy
• Script to run banditP.m

• Fourier Example 1: A known data set
• Fourier Example 2: Sunspots
• The Star Data
• The Letter E data
• Run the Letter E example

1. Review Questions, Exam 1
2. Review Solutions

1. Iris Data

1. Labels for Data below (labels.m) When you run labels.m, you will create a cell structure with names (this is actually a short script file). The names will be stored in "G".
2. Taxonomy Data ( tax001.dat ) This is just a text array of data. Use the load command (like you did for the iris data).
3. Plot routine: plotcell.m
   Useage: once the network "net" is trained using data set X, and assuming you have run labels.m (above), then:
   plotcell(net,X,G)

1. Mushroom data as a Matlab file. Both the inputs (matrix X) and the targets (vector Y) are included. Right click with your mouse, and select "Save Link As..."

1. Matlab file containing the alphabet (as a 35 x 26 matrix) Right click with your mouse, and select "Save Link As..."
2. Function to map a 35-dimensional vector back to a 5 x 7 grid (to see what the letters look like!)

Links to Data For Projects: