2. Content of the Course

Total discipline workload is 32 credit hours.

3. Assessment

Team projects

Examples of the topics:

1.    Deep kernels and Gaussian processes

2.    Bayesian Active Learning

3.    Bayesian black-box optimization

4.    Multi-Fidelity Gaussian Process regression

5.    Bayesian change-point detection

6.    Comparison of approaches for approximation of intractable Bayesian models

7.    MCMC for Bayesian inference

8.    Various applied problems with usage of Bayesian ML methods

Final course project (groups up to 3):

·      Default project topics will be announced on week 2

·      Stages: Project proposal (week 2-3) 

·      Presentation and Final Report submission （week 4）

Final Project types

·      Applied: pick an interesting application and figure out how to apply machine learning algorithms to solve it;

·      Algorithmic: propose a new learning algorithm, or a variant of some existing one to solve a general problem or group thereof.

The Final Report is a PDF:

·      Introduction: motivation and problem statement

·      Related work and brief literature overview

·      Dataset Description

·      ML Methods and algorithms, proposed algorithm modifications, etc.

·      Experiments/Discussion: details about (hyper)parameters and how you picked them, cross-validation metrics and details, discussion of failures and successes, equations, results, visualizations, tables, etc.

·      Conclusion, references, acknowledgements and contributions

Ø  up to 5 pages including figures, tables, appendices (in algorithmicprojects only), and excluding references/contributions

Ø  source code (scripts, notebooks) in ZIP or on Github

The main assessment criteria:

·      General evaluation criteria for the Report

  o  significance, novelty: toy/real problem or common/unexplored method 

  o  technical quality: insightful choice of clever reasonable methods, cross-validation and general quality assessment of used tools/methods

  o  general report quality and structure

  o  relevance to the topics covered during the course

·      The Project presentation

  o  presentation quality and clarity

  o  relevant technical content and summary

  o  the knowledge demonstrated by the team

4. References

Course materials/Textbooks:

• ➜  Bishop, C.M. Pattern Recognition and Machine Learning, Springer, 2007 

• https://www.microsoft.com/en-us/research/people/cmbishop/  

• 
  

• ➜  Barber, D. Bayesian Reasoning and Machine Learning, Cambridge University Press, 2012 

• http://web4.cs.ucl.ac.uk/staff/D.Barber/pmwiki/pmwiki.php?n=Brml.HomePage

• 
  

• ➜  Rasmussen, C., and Williams, C. Gaussian Processes for Machine Learning. The MIT Press, 2006. 

• http://gaussianprocess.org/gpml/

• 
  

• ➜  D. J. C. MacKay. Information Theory, Inference, and Learning Algorithms, 2003. 

  http://wol.ra.phy.cam.ac.uk/mackay/itila/book.html

Name:  Evgeny Burnaev

Email:   e.burnaev@skoltech.ru

            burnaevevgeny@gmail.com

Website:  https://www.researchgate.net/profile/Evgeny_Burnaev

                https://arxiv.org/a/burnaev_e_1.html

                   http://faculty.skoltech.ru/people/evgenyburnaev

                   http://adase.group/