Class number 054–39482R
Class meetings: MW, 9:30am-12:30pm, KAP 147.

Information on this and related pages changes frequently.

Instructor: Sergey Lototsky.

Office: KAP 248D.
Phone: 213 740 2389
E-mail: lototsky usc edu.

Office Hours: MW  after the class. Appointments at other time are welcome.

Course objective:  to learn basic tools for deriving explicit non-asymptotic exponential bounds on probabilities of certain events and to explore some applications of such bounds.

More general goal: to learn something interesting, new, and/or useful.

Official grading scheme: 20% class participation, 40% homework assignments, 40% final presentation.

Main reference: Bernard Bercu, Bernard Delyon, and Emmanuel Rio. Concentration inequalities for sums and martingales. SpringerBriefs in Mathematics. Springer, Cham, 2015. x+120 pp.  ISBN: 978-3-319-22098-7; 978-3-319-22099-4

The book is available in electronic form from the USC Libraries

A longer list of references

Some names and faces 

Homework problems and more

My notes

Other notes

Our Progress

May 21:  Course overview; Elementary inequalities in probability; Some bounds on the normal tail.

May 26: No class (Memorial Day).

May 28: Inequalities of Bernshtein and Hoeffding.

June 2: Sub-Gaussian random variables and Orlicz spaces.

June 4: Gamma and sub-Gamma random variables; conditional expectation and McDiarmid’s inequality.

June 9: Roots of random (Kac) polynomials; martingales and the Azuma-Hoeffding inequality.

June 11: Other martingale inequalities.

June 16: Applications from the book (statistical inference in ARMA models, combinatorial CLT, random matrices).

June 18: Concentration of measure from the analysis point of view (Sobolev, log Sobolev, Poincare, and isoperimetric inequalities, as well as Levy on the sphere and convex Talagrand).

June 23: Hanson-Wright inequality; intro to coupling and related ideas; the Efron-Stein inequality, with applications to LCS and KDE, and entropy extensions.

June 25: From the Boltzmann entropy to KL divergence, Pinsker’s inequality, and optimal transport.

June 30: From concentration inequalities to “with high probability” statements; a first look at uniform limit theorems and VC theory.

July 2: Dimension reduction (Dvoretzky-Milman and Johnsson-Lindenstrauss); matrix versions of Bernshtein, Hoeffding, and Cramer-Chernoff.

July 7: Concluding discussion.