Foundations of data science COM-406


Instructor Ruediger Urbanke
Office INR 116
Email [email protected]
Office Hours By appointment
Teaching Assistant Kirill Ivanov       Office INR 030
Admin Assistant Muriel Bardet       Office INR 137
Lectures Monday 09:15 – 11:00 Room: INM200
  Friday 08:15 – 10:00 Room: INM201
Exercises Friday 10:15 – 12:00 Room: INM201
Language:   English
Credits :   6 ECTS

Lecture notes (PDF file) [these notes will change substantially for the first few weeks; this link will always point to the most recent version]

Official Prerequisites: COM-300 Modèles stochastiques pour les communications (or equivalent)

Here is a link to official coursebook information.

Grading: If you do not hand in your final exam your overall grade will be NA. Otherwise, your grade will be determined based on the following weighted average: 10 % for the Homework, 90 % for the Final Exam.

Course: As long as we are allowed, there will be a lecture in the class and you are welcome to attend. All lectures are recorded and uploaded to the course SWITCHtube channel as soon as possible. In addition, we also do zoom live broadcast of the lectures. 

SWITCHtube channel: https://mediaspace.epfl.ch/channel//29656

Exercises: As long as we are allowed, we will be physically attending the exercise session and you are welcome to attend. There is a live zoom session for those who cannot attend. 

Zoom link: https://epfl.zoom.us/j/95827964653

Update: due to the current situation, lectures go online from Monday Oct. 26. At usual lecture time, there will be live zoom broadcast, which will be recorded and later posted on this page.

Final Exam: The final exam is scheduled on Friday January 29 from 8:15 till 11:15 in CM1121. The exam is open book. You can bring your hand-written notes as well as the lecture notes and use them during the exam. No electronic equipment of any form is allowed (calculator, phone, computer). We will have a Q&A session on Tuesday January 26th from 8:15-10 using our standard zoom link. If you have any questions before that just email us. To help you prepare, here are the exams (with and without solutions for the previous three instances of this course). Good luck!

When Lecture Exercise
Sep 14 introduction (l 1.1), review of probability (l 1.2)  
Sep 18

information measures (l 2.1, l 2.2)

hw 1 (this is a graded homework; submit via moodle) 
Sep 25

information measures (l 3.1, l 3.2)

hw2
Sep 28 information measures, signal representations (l 4.1, l 4.2)  
Oct 2 signal representations (l 5.1, l 5.2)  
Oct 5 signal representations (l 6.1, l 6.2)  
Oct 9

signal representations (l 7.1, l 7.2)

hw3 (this is a graded homework; submit via moodle before Monday Oct.19, 23:59) 
Oct 12 signal representations, estimation (l 8.1, l 8.2)  
Oct 16 estimation (l 9.1, l 9.2)  hw4
Oct 19 detection (l 10.1, l 10.2)  
Oct 23

multi-arm bandits explore-then-exploit

UCB algorithm

 
Oct 26 multi-arm bandits (l 11)  
Oct 30 multi-arm bandits (l 12) hw5 (this is a graded homework; submit via moodle before Monday Nov. 9, 23:59) 
Nov 2 multi-arm bandits, distribution estimation (l 13  
Nov 6 distribution estimation (l 14)  
Nov 9 distribution estimation (l 15)  
Nov 13 distribution estimation (l 16) hw6 
Nov 16 property estimation, exponential family (l 17)  
Nov 20 exponential family (l 18) hw7 (this is a graded homework; submit via moodle before Monday Nov. 30, 23:59) 
Nov 23 exponential family (l 19)  
Nov 27 exponential family (l 20)  
Nov 30 compression (l 21)  
Dec 4 compression (l 22) hw8 (this is a graded homework; submit via moodle before Monday Dec. 14, 23:59) 
Dec 7 compression (l 23)  
Dec 11 exploration bias and generalization bound (l 24)  
Dec 14 generalization bound (l 25)  
Dec 18 review session (l 26)