Sept 15: A1

Sept 22:

A2

Second Assignment:

Write a MATLAB program to implement the MIN-CONFLICTS algorithm to solve the n-queens problem. Find a solution for n=8 and n=15. How many iterations are necessary to find a solution? Plot a histogram of these numbers. What distribution does this histogram resemble?

Due, Sept 29. Please print results and your program listing. Also send program to prof4150@cs.dal.ca with subject line AIA2.

Sept 29:

A3

Given is a training set train1.mat with x and y values, and a test set in file test1.mat with x values.

1. Make a hypothesis (parameterized function) about the function that generated the data in the training set.
2. Specify an objective (cost) function and explain your choice.
3. Fit the function (hypothesis) to the data by using a gradient descent method. Comment on your results.
4. Generate predictions for the x values in the test set.

Due, Tuesday Oct 6 before class so that we can discuss the answers (e.g. no extension possible). Please send your answers, the program and the prediction set to prof4150@cs.dal.ca with subject line AIA3. 

Note: The y values of the training set have imaginary parts. Your prediction (model) could be either for the imaginary numbers or for the real part only.

Oct 6:

A4

Given is a training set train2.mat with x (2 dim) and y values, and a test set in file test2.mat with x values for a binary classification problem.

Generate predictions for the x values in the test set and explain briefly your approach.

Due, Tuesday Oct 13 before class so that we can discuss the answers (e.g. no extension possible). Please send your answers, the program and the prediction set to prof4150@cs.dal.ca with subject line AIA4.

Oct: 13

A5

1. Repeat Assignment 4 with a SVM. You can use a library implementation of a SVM, such as LIBSVM. Explain your choice of parameters.

2. Write a brief proposal (not more than a page) for your research project with your partner.

Course project

The course project will be done in randomly assigned teams. The purpose of the project is to demonstrate and discuss the application of machine learning algorithms, in particular supervised classification, for a specific application. You can choose a data set from the UCI machine learning repository or propose your own application. Study your chosen problem with some of the algorithms which we discussed in class. You need to research on possible literature for your specific problem, and you should consult with the course instructor about your progress. The results of your studies have to be written up in form of a scientific paper, following the paper guidelines given below. A brief guideline and tips for writing a scientific paper will be discussed in class. The project must be presented in class at the end of the term.

Paper guidelines

The paper must be submitted by email to prof4150@cs.dal.ca before Dec 7. 2009, with the subject line `AIproject'. Please use letter format with font size of at least 11pt. The page limit is 6 pages including possible appendices, references and figures. The paper must include a title, the name and affiliation of the authors, and an abstract not exceeding 100 words. References must comply to APA style.

Nov: 24

A6

Implement the MDP example on the 3x4 grid of Russel and Norvig (Figure 1a in script section 11). Implement a reinforcement learning algorithm in Matlab and calculate the optimal value function and the optimal policy. Explain your algorithm and the parameters you chose. Plot the average path length to reach the goal state and the percentage of trials to reach the penelty state during learning.

Due, Tuesday Dec 1. Please send your answers and the executable and commented program to prof4150@cs.dal.ca with subject line AIA6.