Unsupervised learning
Description
Chargé de cours : Nicolas Jouvin
Horaires : Mercredi 8h30-11h45.
News
Exam dates and room: on Friday the 12th of January, 14h - 17h15.
- on paper
- no documents or machines allowed
- annal from last year download here (careful: I was not in charge of the course)
Up-to-date version of the slides: here (there will be frequent updates, keep up to date !)
Outline
- Introduction to Bayesian statistics
- Clustering with finite mixture models
- The EM algorithm
- Hidden Markov Models
- Stochastic Block Model and introduction to variational inference
Séances
- 1ère séance : 3h class + Dowload TD1 exercice sheet + Bonus practical session
- 2eme séance : Expectation-Maximization algorithm. Download the TD2 exercise sheet.
- 3ème séance : Introduction to HMMs + Bonus practical session on discrete Markov chains
- 4ème séance : Inference in HMMs : Forward-Backward & EM (aka Baum-Welch)
- 5ème séance : Viterbi algorithm and introduction to SBM + Practical session on HMMs: EM and Viterbi
- 6ème séance : Variational inference & illustration with SBM
Corrections
On the fly version of the code during TD on GMM downloadable here