Advanced Course in Machine Learning : Exercises

The exercises are revealed every Tuesday and are due next week's Tuesday. The solutions should be returned by email to the course organizers -- see the exercise paper for more details. Model solutions for the exercises are presented in the following exercise session on Wednesday and put avaiable here around the same time.

For instructions on retrurning the exercises see the first page of the problem set. Note that the pen&paper exercises can sitll be returned during the lecture on the due date.

Many of the exercise problems may seem a bit ambiguous or lacking practical details. This is typically intentional: The point is that you should stop and think about how exactly the problem could be solved. For example, the exercises might say "Run the algorithm until convergence", without specifying how exactly convergence is defined. Instead of giving up, you should choose some criterion and also explain your choice as part of the solution ("To determine convergence, I did ..., and the result was...")

 

The bonus exercise: The eight set of exercises is voluntary and intended for compensating some earlier exercise sessions you could not return, or simply to collect more points to improve the grade. The points for the bonus exercise will be added for your total, but you do not gain anything by getting more than 7*24 points for all of the exercises.

 

Note: Two copies of the course book will be available in the exercise sessions.

Note: The exercises below refer to the course in 2016. Access of the model solutions have now been removed since some of them might be re-used in the 2017 course.  

 

Problems Released Due Model solutions Additional links
Exercise 1 March 15 March 22 Solutions 1 Data set for problem 6 (NOTE: The file contains transpose of X)
Exercise 2 March 18 April 5 Solutions 2, code for poblem 2, code for problem 3 Data for problem 2: train, test  /   Data for problem 3: train
Exercise 3 April 1 April 12 Solutions 3 (fixed version May 6), code for problem 3 Data for problem 3
Exercise 4 April 8 (fixed version April 15) April 19 Solutions 4, code for problem 3, code for problem 4 Data for problem 3 / Data for problem 4 (from the Introduction to ML course)
Exercise 5 April 15 April 26 Solutions 5  
Exercise 6 April 22 May 3 Solutions 6, code for problem 1  
Exercise 7 April 29 May 17 Solutions 7, codes Data for problem 2: train, test 
Bonus exercise May 3 May 17 Solutions, code for problem 1 Data for problem 1