WARNING: This is the webpage of the 2019-2020 course!
Please check: http://caio-corro.fr/teaching/
Lab exercises

Lab exercises

There should be 1 or 2 students per group. Deadlines are hard and you cannot change groups. Groups can be different from paper presentation and project groups.

27 nov. 2019
11:15 - 12:15


Lab exercise 1

download notebook / download dataset / view as html

Hard deadline: December 3, 2019.

4 dec. 2019
11:15 - 12:15


18 dec. 2019
12:15 - 13:15


Lab exercise 2

The goal is to extend the first lab exercise by using a convolutional neural network instead of a bag-of-word model. Of course, you have to build it youtself by creating your custom CNN module, you can not use the convolutional layers provided by Pytorch. For this lab exercise, you need to explain really clearly what is going on in your code (comments!) and to describe precisely the architecture in the report, with your own notation.

Bonus: also try with a LSTM classifier! (only if you have enough time)

To understand what is a convolutional neural network:

Hard deadline: December 22, 2019.