Thematic CERN School of Computing on Machine Learning 2024

13 Oct 2024, 15:00 → 19 Oct 2024, 11:00 Europe/Zurich

MedILS, Split, Croatia

Alberto Pace (CERN), Andrzej Nowicki (CERN), Kristina Gunne (CERN)

The 15th Thematic CERN School of Computing (tCSC Machine Learning 2024) will take place on October 13-19, 2024. 

The school will focus on the theme of Machine Learning and Artificial Intelligence applied to Data Analysis and Accelerator Technology. The programme will offer 22 hours of lectures and hands-on exercises, and student presentation sessions.

This school is organized by CERN in collaboration with the Faculty of Science, University of Split. The school will take place in Split, Croatia, and be hosted at the Mediterranean Institute For Life Sciences (MEDILS) Conference Centre. The Centre is a historical renovated building situated in a wooded and landscaped park located on the Adriatic Sea coast, a few kilometers from the centre of Split.

Important dates 2024

• 8 May application opens
• 19 June application close
• 3 July invitations sent to selected students
• 4 September participation fee deadline

 

Sunday, 13 October

1 Registration at MedILS

2 Welcome and self presentation session

Speakers: Alberto Pace (CERN), Andrzej Nowicki (CERN), Kristina Gunne (CERN)

3 Transport to Split

4 Guided tour of Split

19:30 Welcome dinner

Monday, 14 October

5 opening session

Speaker: Alberto Pace (CERN)

6 Machine learning methods: L1 Introduction to Statistics

In this lecture we will go over key concepts in statistics which are the cornerstone of mathematical foundation of Machine Learning. We will define frequentistic and Bayesian probabilities, learn what is a PDF. We will also discuss parameter estimation with the Maximum Likelihood method and finish with the definition of Confidence Intervals.

Speakers: Francesco Vaselli (Scuola Normale Superiore & INFN Pisa (IT)), Toni Sculac (University of Split Faculty of Science (HR))

7 Announcements

11:00 Coffee

8 Machine learning methods: L2 Statistics and Machine Learning

We start this lecture with unfolding and hypothesis testing, another two key concepts from statistics. Key part of the lecture is the Neyman-Person lemma that paves a clear path for the needs of Machine Learning in statistics.

Speakers: Francesco Vaselli (Scuola Normale Superiore & INFN Pisa (IT)), Toni Sculac (University of Split Faculty of Science (HR))

12:30 Lunch

9 Study time or daily sports

15:15 Coffee

10 Machine learning methods: L3 Classical Machine Learning

We continue tackling the problem of trying to know the likelihood ratio with the use of Classical Machine Learning. We try to solve it by brute force and then we move to Machine Learning techniques. We start with a Kernel Density Estimators. We continue by defining what is a decision tree, what is a leaf and we study how it works on a very simple example. We go further and explain the difference between classification and regression, as well as the need for pruning, bagging, and boosting. This main goal of this lecture is to remove the idea of the “black-box approach" and understand all of the details of a decision tree.

Speakers: Francesco Vaselli (Scuola Normale Superiore & INFN Pisa (IT)), Toni Sculac (University of Split Faculty of Science (HR))

16:45 Break

11 Machine Learning methods: excercise 1

Speakers: Francesco Vaselli (Scuola Normale Superiore & INFN Pisa (IT)), Toni Sculac (University of Split Faculty of Science (HR))

12 Machine Learning methods: excercise 2

Speakers: Francesco Vaselli (Scuola Normale Superiore & INFN Pisa (IT)), Toni Sculac (University of Split Faculty of Science (HR))

19:30 Dinner at MedILS

Tuesday, 15 October

13 Machine Learning in Accelerator Technologies: Machine Learning for particle accelerators

Main use cases and applications

Speaker: Verena Kain (CERN)

14 Machine Learning in Accelerator Technologies: Bayesian Optimisation

Speaker: Michael Schenk (CERN)

15 Announcements

11:00 Coffee

16 Machine Learning Methods: L4 Introduction to Deep Learning

We introduce the concept of a Neural Network (NN) and study their application with a single-neuron network. This again allows us to avoid the "black-box approach" and really understand the key concepts of how a NN works. We discuss activation functions and how the NN learns with the help of the loss functions and backpropagation. We finish by discussing the basic idea of a Deep Neural Network and basic training concepts.

Speakers: Francesco Vaselli (Scuola Normale Superiore & INFN Pisa (IT)), Toni Sculac (University of Split Faculty of Science (HR))

12:30 Lunch

17 Study time or daily sports

15:15 Coffee

18 Machine Learning methods: exercise 3

Speakers: Francesco Vaselli (Scuola Normale Superiore & INFN Pisa (IT)), Toni Sculac (University of Split Faculty of Science (HR))

16:45 Break

19 Machine learning in accelerators: Exercise 1

20 Machine learning in accelerators: Exercise 2

19:30 Dinner at MedILS

Wednesday, 16 October

21 Machine Learning in Accelerators: Introduction to Reinforcement Learning

22 Machine Learning in Accelerators: Advanced concepts for Reinforcement Learning

23 Announcements

11:00 Coffee

24 Machine learning in accelerators: Exercise 3

12:15 Lunch

25 River rafting excursion

Departure from MedILS at 13h

18:30 Dinner at Kastel Slanica Omis

26 Transport back to medILS

Thursday, 17 October

27 Machine learning in Data Analysis: Introduction to Machine Learning for HEP, Anomaly detection and real time applications

Speaker: Dr Sofia Vallecorsa (CERN)

28 Machine learning in Data Analysis: The data reconstruction step - a pattern recognition problem

Speaker: Dr Sofia Vallecorsa (CERN)

29 Announcements

11:00 Coffee

30 Machine learning in Data Analysis: Generative Models for HEP

12:30 Lunch

31 Study time or daily sports

15:15 Coffee

32 Machine learning in Data Analysis: Exercise 1

16:45 Break

33 Machine learning in Data Analysis: Exercise 2

34 Machine learning in Data Analysis: Exercise 3

19:30 Dinner

Friday, 18 October

35 Lightning talks

36 Machine learning in Data Analysis: Systematics in ML

Speaker: Dr Sofia Vallecorsa (CERN)

37 Announcements

11:00 Coffee

38 Machine Learning Methods: L5 Advanced Deep Learning

Speakers: Francesco Vaselli (Scuola Normale Superiore & INFN Pisa (IT)), Toni Sculac (University of Split Faculty of Science (HR))

12:30 Lunch

39 Exam

14:30 Break

40 Closing ceremony

Speaker: Alberto Pace (CERN)

41 Sports and leisure time

19:30 Closing dinner

Saturday, 19 October

42 Departures from MedILS