From Basic Machine Learning models to Advanced Kernel Learning


Course description

Statistical learning is about the construction and study of systems that can automatically learn from data. With the emergence of massive datasets commonly encountered today, the need for powerful machine learning is of acute importance. Examples of successful applications include effective web search, anti-spam software, computer vision, robotics, practical speech recognition, and a deeper understanding of the human genome.

This course gives an introduction to this exciting field. In the first part, we will introduce basic techniques such as logistic regression, multilayer perceptrons, nearest neighbor approaches, both from a theoretical and methodological point of views. In the second part, we will focus on more advanced techniques such as kernel methods, which is a versatile tool to represent data, in combination with (un)supervised learning techniques that are agnostic to the type of data that is learned from. The learning techniques that will be covered include regression, classification, clustering and dimension reduction. We will cover both the theoretical underpinnings of kernels, as well as a series of kernels that are important in practical applications. Finally we will touch upon topics of active research, such as large-scale kernel methods and the use of kernel methods to develop theoretical foundations of deep learning models.


Evaluation

The grading of the class will be done with (i) one final exam, (ii) two homeworks.

Reading material

Schedule

Lectures are scheduled from 9:45-11:15 on Mondays, and from 8:15-9:45 on Thursdays. The first part of the class will recap basic supervised learning techniques, theory, and algorithms. Lecture notes or slides will be updated here on the fly.

You can download the slides for all lectures of the second part of the class (advanced kernel methods) here! Each lecture corresponds to a range of slides. Slides are frequently updated. Please let us know if you spot typos!

Homeworks. The first homework (data) is due by Monday, November 21, 2022. It is to be send by email to Pierre Gaillard as a pdf file containing all results and figures, together with the code file (e.g., notebook, python file,...). The report can be written in English or in French.

# Date Room Teacher Topic
1 26/09/2022 H102 PG Supervised learning basics
2 29/09/2022 H204 PG Linear regression
3 03/10/2022 D117 PG Linear regression
4 06/10/2022 H206 PG Logistic regression
5 10/10/2022 H203 PG Logistic regression and Maximum likelihood
6 17/10/2022 H105 PG Maximum likelihood and K-Nearest Neighbors
7 20/10/2022 H204 PG K-Nearest Neighbors (notebook)
8 24/10/2022 H104 PG Multilayer Perceptrons
9 27/10/2022 H204 MA Positive definite kernels, slides 1-23 (Homework 1)
10 7/11/2022 H103 MA RKHS I, slides 25-55
11 10/11/2022 H206 MA RKHS II, slides 25-55
12 14/11/2022 H105 MA Kernel tricks, slides 56-82
13 17/11/2022 D111 MA Kernel Ridge regression, slides 83-103
14 21/11/2022 H103 MA (Homework 1 due date)Large-margin classifiers, slides 115-132, Suplementary material: Rademacher complexity
15 24/11/2022 H103 JM Support Vector Machines, slides 134-165
16 28/11/2022 Video only JM kernel PCA, slides 166-202 Video I
17 01/12/2022 Video only JM kernel K-means, kernel CCA, slides 166-202 Video II
18 05/12/2022 H104 PG Q&A on the videos. Kernel Jungle, slides 203-...
19 08/12/2022 H102 JM
20 12/12/2022 H205 MA
21 15/12/2022 H202 MA
22 02/01/2023 H204 JM
23 05/01/2023 H203 JM
24 12/01/2023 D211 MA
# Week of 23/01 Exam

Instructors

Contact: firstname.lastname@inria.fr