Consultation hours
Send me an email to schedule an appointment online.
Course
Modern Methods of Decision Making
Master Program
-
Data Science
Meetings
-
Mondays, from 11:10 to 14:20, from January 11 to June 14, 2021
Format
-
Online
-
Zoom link: here
Main references for the course:
-
S. Shalev-Shwartz and S. Ben-David (2014). Understanding Machine Learning: From Theory to Algorithms. Cambridge University Press.
-
S. Bubeck (2015). Convex Optimization: Algorithms and Complexity. Foundations and Trends in Machine Learning. Vol. 8, No. 3-4.
Home assignments
-
Home assignment 1: here (due February 26)
-
Home assignment 2: coming soon
Previously:
-
Video recordings: here
-
Lecture 0: Introduction (January 11)
-
Lecture 1: Statistical vs. online learning (January 18).
-
Notes: here
-
-
Lecture 2: Tools from probability theory (January 25)
-
Lecture 3: Recap on linear algebra and differential calculus (February 1st)
-
Lecture 4: Convexity (February 8)
-
Lecture 5: Empirical risk minimisation I (Feb 15)
-
Lecture 6: Empirical risk minimisation II (March 1)
-
Notes: here
-
No seminar
-
Next:
-
Lecture 7: Support vector machines (March 8)
-
Lecture 8: Boosting
-
Lecture 9: Gradient descent
-
Lecture 10: Mirror descent
-
Lecture 11: Stochastic optimisation
-
Lecture 12: Introduction to online learning
-
Lecture 13: Prediction with expert advice
-
Lecture 14: First order methods for online convex optimisation
-
Lecture 15: Online newton step algorithm
-
Lecture 16: Stochastic bandit algorithms
-
Lecture 17: Adversarial bandit algorithms
-
Lecture 18: Bandit convex optimisation
Course
High dimensional probability and statistics
Master Program
-
Statistical Learning Theory
Meetings
-
Wednesdays, from 13:00 to 16:00, from January 27 to March 17, 2021
Format
-
Online
-
Zoom link: here
Main references for the course:
-
R. Vershynin (2018). High-Dimensional Probability. An Introduction with Applications in Data Science. Cambridge University Press.
-
M. Wainwright (2019). High-Dimensional Statistics. A Non-Asymptotic Viewpoint. Cambridge University Press.
Home assignments
-
First: here (due February 24)
Previously:
-
Video recordings: here
-
Lecture 1: Concentration (January 27)
-
Notes: here
-
-
Lecture 2: Sums of independent random variables (February 3)
-
Notes: here
-
-
Lectures 3 & 4: Suprema (February 10 & 17)
-
Notes: here
-
-
Lecture 5: The Johnson-Lindenstrauss lemma (February 24)
-
Notes: here
-
-
Lecture 6: Cov. matrix estimation and PCA (March 3)
-
Notes: coming soon
-
Next
-
Lecture 7: Concentration of random matrices (March 10)
-
Lecture 8: Community detection in random graphs (March 17)
-
Lecture 9: High-dimensional linear regression (March 24)
Course
Gradients flows in metric spaces
PhD Program
Meetings
-
Thursdays, from 18:10 to 21:00, from January 14 to March 18, 2021
Format
-
Online
-
Zoom link: here
References for the course
-
S. Danieri and G. Savaré (2014). Lecture notes on gradient flows and optimal transport. In Y. Ollivier, H. Pajot, & C. Villani (Eds.), Optimal Transport: Theory and Applications (London Mathematical Society Lecture Note Series, pp. 100-144). Cambridge: Cambridge University Press. (available on arxiv here)
-
L. Ambrosio, N. Gigli and G. Savaré (2005). Gradient Flows. Birkhauser.
-
L. Ambrosio and N. Gigli (2009). A user’s guide to optimal transport. (available here)
Previously
-
Lecture 0: Intro
-
Lecture 1: Review of gradient flows in Euclidean spaces
-
Notes: here
-
-
Lecture 2: Definitions of gradient flows in metric spaces: EDI, EDE and EVI
-
Notes: here
-
-
Lecture 3: Minimizing movement scheme and existence of EDI gradient flows.
-
Notes: coming soon
-
Next
-
Lecture 4