2019/2020

***Instructors:Prof. Nicolas VAYATIS

Teaching assistant: Xavier FONTAINE

***Emails:

<name/-at-/cmla.ens-cachan.fr>

***Syllabus 2019-2020

***Course schedule and location:

Room Condorcet, Building d'Alembert, ENS Paris-Saclay (Cachan campus - 61, avenue du Président Wilson - 94230 Cachan)

Date | Time | Instructor | Session | Topics | Files/Readings |

Tuesday October 1 |
11:00-13:00 | N. Vayatis | Lecture #1 | Course introduction and main setup Chapter 1 - Optimality in statistical learning |
Slides |

Tuesday October 8 |
11:00-13:00 | N. Vayatis | Lecture #2 | Chapter 1 - Optimality in statistical learning |
Slides |

Tuesday October 15 |
11:00-13:00 | X. Fontaine | Exercise session #1 | Optimal elements and excess risk bounds |
Sheet |

Tuesday October 22 |
11:00-13:00 | N. Vayatis | Lecture #3 | Chapter 2 - Probabilistic inequalities, complexity
measures |
Slides |

Tuesday October 29 |
11:00-13:00 | X. Fontaine | Exercise session #2 | Inequalities, Rademacher complexity, VC dimension |
Sheet |

Tuesday November 5 |
11:00-13:00 | Partial exam - MandatoryNo documents allowed |
|||

Tuesday November 12 |
11:00-13:00 | N. Vayatis | Lecture #4 | Chapter 2 - Regularization, stability |
Slides |

Tuesday November 19 |
11:00-13:00 | N. Vayatis | Lecture #5 | Chapter 3 - Consistency of Machine Learning methods
(I) Boosting |
Slides |

Tuesday November 26 | 11:00-13:00 | X. Fontaine | Exercise session #3 | Consistency and convergence bounds |
Sheet |

Tuesday December 3 |
11:00-13:00 | N. Vayatis | Lecture #6 | Chapter 3 - Consistency of Machine Learning methods
(II) SVM, Feedforward NN |
Slides |

Tuesday December 10 | 11:00-13:00 | X. Fontaine | Exercise session #4 | Course wrap-up |
Sheet |

Wednesday January 8 | 13:30-15:30 | Final examDocuments allowed |
Rooms C406 and C415 - Cournot
building 4th floor |

***Office hours:

Tuesdays 1pm/2pm - Laplace 126

***Evaluation:

Partial exam on November 5 (am) - MANDATORY

Final exam on January 8 (1:30pm-3:30pm) - MANDATORY

***Grading:

Course grade = max(final ; (partial+final)/2)

*************

From information theory to statistical learning

The probabilistic view on numerical data,

Optimality and the bias-variance dilemma

Chapter 2 - The mathematical foundations of statistical learning:

Risk minimization

Concentration inequalities

Complexity measures

Regularization

(Stability)

Chapter 3 - Theory: Consistency theorems and error bounds of learning algorithms

Part 1 - SVM, Neural Networks, Boosting

Part 2 - Bagging, Random Forests

(Chapter 4 - Advanced topics)

Disclaimer: this document was offered by a former student, some mistakes may be found in the document.

References

- S. Boucheron, O. Bousquet, and G. Lugosi. Theory of Classification: a Survey of Recent Advances. ESAIM: Probability and Statistics, 9:323375, 2005.
- L. Devroye, L. Györfi, G. Lugosi, A Probabilistic Theory
of Pattern Recognition, Springer, New York,
*1996.* - G. Lugosi,
*Principles of Nonparametric Learning*Springer, Wien, New York, pp. 1--56, 2002. - M. Mohri, A. Rostamizadeh, A. Talwalkar. Foundations of Machine Learning, The MIT Press, 2012.
- S. Shalev-Schwartz, S. Ben-David. Understanding Machine
Learning: From Theory to Algorithms.Cambridge University
Press, 2014.