Ensimag Rubrique Formation 2022

Model selection for large-scale learning - WMM9AM24

  • Number of hours

    • Lectures 15.0
    • Projects -
    • Tutorials -
    • Internship -
    • Laboratory works 3.0
    • Written tests -

    ECTS

    ECTS 3.0

Goal(s)

We will focus on the penalized empirical risk, where the penalty may be deterministic (as BIC or ICL) or estimated with data (as the slope heuristic).
The goal is to understand why model selection may be useful and to know the basic criterion to do it.

Responsible(s)

Emilie DEVIJVER

Content(s)

When estimating parameters in a statistical model, sharp calibration is important to get optimal performances. In this course, we will focus on the selection of estimators with respect to the data. Particularly, we will consider calibration of parameters (e.g., regularization parameter for minimization of regularized empirical risk, like Lasso or Ridge estimators) and model selection (where each estimator minimizes the empirical risk on a specified model, as mixture models with several number of clusters).

Prerequisites

Basic knowledges in probability and statistics

Test

CC : reading of a research paper
Exam1 : test
Exam 2 : test or oral

N1 = 50% CC + 50% Exam1
N2 = 100% Exam2

The exam is given in english only FR

Calendar

The course exists in the following branches:

  • Curriculum - Master 2 in Applied Mathematics - Semester 9 (this course is given in english only EN)
  • Curriculum - Master 2 in Computer Science - Semester 9 (this course is given in english only EN)
see the course schedule for 2020-2021

Additional Information

Course ID : WMM9AM24
Course language(s): FR

You can find this course among all other courses.