Material related to homework assignments will appear here.

Homework policies

Note: These are tentative guidelines. There may be changes to details of how many points are available for each type of homework etc.

Addition to homework policies: beginning from Exercise set 6, please include with your homework also a short note mentioning which students you worked with, and which external sources of information you used (if any). [added 27 November]

Students are welcome, but not required, to attend the exercise session (Thu 14–16) to discuss the homework problems in small groups. Individual submissions are however required and they should be based on independent work by the student. As a rule of thumb, after discussing the homework, you should do something unrelated for half an hour and then write down your solution by yourself, without using written notes about the discussion.

Help is also available on the IRC channel #tkt-iml.

The submissions should be sent as a single email to Johannes and Amin. Please attach a single pdf and compressed file containing your answers and code respectively. The deadline for submissions is Fridays at 23:59. There is a 8-hour grace period (until Saturday 08:00) for late submissions, which will be penalised in grading.

The pen&paper problems are generally graded with 0–3 points, 3 being a complete solution, 2 missing something and 1 for an honest try. Late submissions (received during the grace period) will have a maximum of 2 points rewarded per question.

The programming problems are graded with 0–15 points, 15 being a complete solution, 10 missing something and 5 for an honest try. Intermediate points are awarded based on style and efficiency. The maximum available points for late submissions (received during the grace period) is 12.

Grading will in general be done during the weekend and students can expect to see their grades by Monday noon. Model answers and grade distributions will be available before the next lecture (Tuesday 10:00) and discussed shortly in the beginning of the Thursday exercise session.

Homework problems

You can find your results using the department's TIKLI service.

• week 1: sample problems and a brief Python tutorial for the Python guidance session
• week 2 (deadline was Friday 6 November): problems, sample solutions
• week 3 (deadline was Friday 13 November): problems, sample solutions
• week 4 (deadline was Friday 20 November): problems, sample solutions
• week 5 (deadline was Friday 27 November): problems, sample solutions
• week 6 (deadline Friday 4 December): problems
• week 7 (deadline Friday 11 December): problems

Data sets

• MovieLens
• MNIST

Links above lead to packages prepared for this course including some helpful routines for handling the data. The original data sets are due to GroupLens Research and Yann LeCun et al.

Solutions