Assignments

All assignments will be due on Tuesdays at the start of class.
Assignments will be posted on this web page at least one week before they are due.

Each of the 5 assignments will be worth 10% of the final grade.
Except in the case of an official Student Medical Certificate, assignments that are submitted late will have their score reduced by 20% of the maximum achievable score per day. The time past the deadline will be rounded up to an integer number of days.

You may discuss the general idea of the questions with anyone you like, but your discussion may not include the specific answers to any of the problems. In other words, you should always be able to answer, by yourself, the question "How did you get this answer?"

A typical assignment will have two parts, both related to the same learning procedure. The first part will require you to answer questions about the learning procedure. The second part will require you to modify some Matlab code that implements a simple version of the learning procedure and describe the results you obtain using the modified code.
Assignment 1 will be based on the backpropagation algorithm.
Assignment 2 will involve using backprop to get good performance on a validation set
Assignment 3 involves fitting mixtures of Gaussians.
Assignment 4 involves fitting Hidden Markov Models.
Assignment 5 involves playing with Restricted Boltzmann Machines.

Tests

The midterm test (worth 25% of the total) is on February 25. It starts at 1.10am and lasts for 45 minutes. It is closed book. You will get your grades the following week before the March 9 deadline for dropping courses.
The final exam (worth 25% of the total) is on April 21 from 2.00 to 4.00. It lasts for TWO HOURS and will be closed book. It only tests material that came after the midterm.

Missed tests will get a score of zero except in the case of an official Student Medical Certificate or a written (not email) request submitted at least one week before the test date and approved by the instructor.