Sixth Machine Learning in High Energy Physics Summer School 2020

Europe/Berlin
Zoom

Zoom

Andrey Ustyuzhanin (Yandex School of Data Analysis (RU)) , Lesya Shchutska (EPFL - Ecole Polytechnique Federale Lausanne (CH))
Description

The Sixth Machine Learning summer school organised by Yandex School of Data AnalysisLaboratory of Methods for Big Data Analysis of National Research University Higher School of Economics, and High Energy Physics Laboratory LPHE at EPFL will be held at EPFL, Lausanne, Switzerland from the 16th to 30th of July 2020.

The school will cover the relatively young area of data analysis and computational research that has started to emerge in High Energy Physics (HEP). It is known by several names including “Multivariate Analysis”, “Neural Networks”, “Classification/Clusterization techniques”. In more generic terms, these techniques belong to the field of “Machine Learning”, which is an area that is based on research performed in Statistics and has received a lot of attention from the Data Science community.

There are plenty of essential problems in high energy physics that can be solved using Machine Learning methods. These vary from online data filtering and reconstruction to offline data analysis.

Students of the school will receive a theoretical and practical introduction to this new field and will be able to apply acquired knowledge to solve their own problems. Topics ranging from decision trees to deep learning and hyperparameter optimisation will be covered with concrete examples and hands-on tutorials. A special data-science competition will be organised within the school to allow participants to get better feeling of real-life ML applications scenarios.

The expected number of students for the school is about 60. The school is aimed at PhD students and postdoctoral researchers, but also open to masters students.

Pre-requisites for participation

Upon completion of the school participants would be able to

  • formulate a HEP-related problem in ML-friendly terms;
  • select quality criteria for a given problem;
  • understand and apply principles of widely-used classification models (e.g. boosting, bagging, BDT, neural networks, etc) to practical cases;
  • optimise features and parameters of a given model in efficient way under given restrictions;
  • select the best classifier implementation amongst a variety of ML libraries (scikit-learn, xgboost, deep learning libraries, etc);
  • understand and apply principles of generative model design;
  • define & conduct reproducible data-driven experiments.

School materials

School materials are available as online course.

Organisers

High Energy Physics Laboratory LPHE at EPFL

Partners

 

Sponsors

Thanks to our sponsors:

CHIPP: Swiss Institute of Particle Physics    Swiss Academy of Sciences  
EPFL Doctoral School
 

 

we can provide a bit of subsidy for students who are not able to afford the full registration fee out of their own funds. Make sure you apply before the early registration deadline. See registration fee page for details. 

    • 1
      Welcome, opening words

      Welcome by organizers, the structure and gist of the school.

      Speaker: Andrey Ustyuzhanin (Yandex School of Data Analysis (RU))
    • 2
      Practice. Getting familiar with school infrastructure
      Speakers: Andrey Ustyuzhanin (Yandex School of Data Analysis (RU)) , Vladislav Belavin (Yandex School of Data Analysis (RU))
    • 10:00 AM
      Break
    • 3
      Intro into Machine Intelligence

      Overview of ML history, applications and outlook with HEP perspective.

      Speaker: Andrey Ustyuzhanin (Yandex School of Data Analysis (RU))
    • Section 1. Introduction into ML: Basics
      Convener: Artem Maevskiy (National Research University Higher School of Economics (RU))
      • 4
        Introduction into supervised learning

        Practical session

        Speaker: Artem Maevskiy (National Research University Higher School of Economics (RU))
      • 5
        Seminar: Data Handling in Python
        Speaker: Artem Maevskiy (National Research University Higher School of Economics (RU))
    • 12:45 PM
      Lunch
    • Section 1. Introduction into ML: Linear Regression
      Convener: Artem Maevskiy (National Research University Higher School of Economics (RU))
      • 6
        Linear Regression

        Linear regression. Analytical solution. Gradient descent. Numerical solution. Polynomial features.

        Speaker: Artem Maevskiy (National Research University Higher School of Economics (RU))
      • 7
        Seminar
        Speaker: Artem Maevskiy (National Research University Higher School of Economics (RU))
    • 2:30 PM
      Break
    • Section 1. Introduction into ML: Logistic regression-1
      Convener: Artem Maevskiy (National Research University Higher School of Economics (RU))
      • 8
        Logistic regression-1
        Speaker: Artem Maevskiy (National Research University Higher School of Economics (RU))
      • 9
        Seminar
    • 4:00 PM
      Break
    • Section 1. Introduction into ML: Logistic regression-2
      Convener: Artem Maevskiy (National Research University Higher School of Economics (RU))
      • 10
        Logistic regression-2

        Linear models regularization. Probabilistic interpretation of linear models (regression and classification).

        Speaker: Artem Maevskiy (National Research University Higher School of Economics (RU))
      • 11
        Seminar
        Speaker: Artem Maevskiy (National Research University Higher School of Economics (RU))
    • Section 1. Introduction into ML: Quality Metrics-1
      Convener: Mikhail Hushchyn (Yandex School of Data Analysis (RU))
      • 12
        Quality Metrics-1

        Quality metrics for classification and regression

        Speaker: Mikhail Hushchyn (Yandex School of Data Analysis (RU))
      • 13
        Seminar: Quality Metrics-1
    • 10:00 AM
      Break
    • Section 1. Introduction into ML: Quality Metrics-2
      Convener: Mikhail Hushchyn (Yandex School of Data Analysis (RU))
      • 14
        Quality Metrics-2

        How to test your model. Cross validation. Statistical model comparison

        Speaker: Mikhail Hushchyn (Yandex School of Data Analysis (RU))
      • 15
        Seminar: Quality Metrics-2
        Speaker: Mikhail Hushchyn (Yandex School of Data Analysis (RU))
    • 11:30 AM
      Break
    • Section 1. Introduction into ML: Decision trees-1
      Convener: Nikita Kazeev (Yandex School of Data Analysis (RU))
      • 16
        Decision trees-1

        Splitting rule. Classification and regression decision trees

        Speaker: Mr Nikita Kazeev (Yandex School of Data Analysis (RU))
      • 17
        Seminar: Decision trees-1
        Speaker: Nikita Kazeev (Yandex School of Data Analysis (RU))
    • 1:00 PM
      Lunch
    • Section 1. Introduction into ML: Ensembles-1
      Convener: Nikita Kazeev (Yandex School of Data Analysis (RU))
      • 18
        Ensembles-1

        Bagging and Random Forest. Stacking and blending.

        Speaker: Nikita Kazeev (Yandex School of Data Analysis (RU))
      • 19
        Seminar: Ensembles-1
        Speaker: Nikita Kazeev (Yandex School of Data Analysis (RU))
    • 3:00 PM
      Break
    • Section 1. Introduction into ML: Ensembles-2
      Convener: Nikita Kazeev (Yandex School of Data Analysis (RU))
      • 20
        Ensembles-2

        Gradient boosting.

        Speaker: Nikita Kazeev (Yandex School of Data Analysis (RU))
      • 21
        Seminar: Ensembles-2
        Speaker: Nikita Kazeev (Yandex School of Data Analysis (RU))
    • 22
      Coopetition introduction (1)
      Speaker: Mr Nikita Kazeev (Yandex School of Data Analysis (RU))
    • 5:00 PM
      Break
    • Socialization: Open flow discussion
    • Section 1. Introduction into ML: Useful hacks.
      Convener: Mikhail Hushchyn (Yandex School of Data Analysis (RU))
      • 23
        Useful hacks

        Feature engineering, importance and selection.

        Speaker: Mikhail Hushchyn (Yandex School of Data Analysis (RU))
      • 24
        Seminar: Useful hacks.
        Speaker: Mikhail Hushchyn (Yandex School of Data Analysis (RU))
    • 10:00 AM
      Break
    • Section 1. Introduction into ML: Clustering-1
      Convener: Mikhail Hushchyn (Yandex School of Data Analysis (RU))
      • 25
        Clustering-1

        Clustering. K-Means. Quality metrics for clustering

        Speaker: Mikhail Hushchyn (Yandex School of Data Analysis (RU))
      • 26
        Seminar: Clustering-1
        Speaker: Mikhail Hushchyn (Yandex School of Data Analysis (RU))
    • 11:30 AM
      Break
    • Section 1. Introduction into ML: Clustering-2
      Convener: Mikhail Hushchyn (Yandex School of Data Analysis (RU))
      • 27
        Clustering-2

        Hierarchical clustering and DBSCAN.

        Speaker: Mikhail Hushchyn (Yandex School of Data Analysis (RU))
      • 28
        Seminar: Clustering-2
        Speaker: Mikhail Hushchyn (Yandex School of Data Analysis (RU))
    • 9:00 AM
      Sunday
    • Section 2. Introduction into Neural Networks: Intro to NN
      Convener: Artem Maevskiy (National Research University Higher School of Economics (RU))
      • 29
        Intro to NN
        Speaker: Artem Maevskiy (National Research University Higher School of Economics (RU))
      • 30
        Seminar: Intro to NN
        Speaker: Artem Maevskiy (National Research University Higher School of Economics (RU))
    • 10:00 AM
      Break
    • Section 2. Introduction into Neural Networks: Intro to Pytorch
      Convener: Andrey Ustyuzhanin (Yandex School of Data Analysis (RU))
      • 31
        Intro to Pytorch
        Speaker: Andrey Ustyuzhanin (Yandex School of Data Analysis (RU))
      • 32
        Seminar: Pytorch practice
        Speaker: Andrey Ustyuzhanin (Yandex School of Data Analysis (RU))
    • 11:30 AM
      Break
    • Section 2. Introduction into Neural Networks: CNN
      Convener: Andrey Ustyuzhanin (Yandex School of Data Analysis (RU))
      • 33
        CNN
        Speaker: Andrey Ustyuzhanin (Yandex School of Data Analysis (RU))
      • 34
        Seminar: CNN
        Speaker: Andrey Ustyuzhanin (Yandex School of Data Analysis (RU))
    • 1:00 PM
      Lunch
    • Section 2. Introduction into Neural Networks: Network regularization
      Convener: Artem Maevskiy (National Research University Higher School of Economics (RU))
      • 35
        Network regularization
      • 36
        Seminar: Network regularization
    • 3:00 PM
      Break
    • Section 2. Introduction to Neural Nets: Autoregressive networks
      Convener: Artem Ryzhikov (National Research University Higher School of Economics (RU))
      • 37
        Autoregressive networks
      • 38
        Seminar: Autoregressive networks
    • 4:30 PM
      Break
    • Guest lectures
      Convener: Tommaso Dorigo (Universita e INFN, Padova (IT))
    • Section 2. Introduction to Neural Nets: Autoregressive networks-2
      Convener: Artem Ryzhikov (National Research University Higher School of Economics (RU))
      • 39
        Autoregressive networks-2
      • 40
        Seminar: Autoregressive networks-2
    • 10:00 AM
      Break
    • Section 2. Introduction to Neural Nets: Network architectures: tips and tricks
      Convener: Maxim Borisyak (Yandex School of Data Analysis (RU))
      • 41
        Network architectures: tips and tricks
      • 42
        Seminar: Practice
    • 11:30 AM
      Break
    • Section 3. Bayesian Deep Learning: Intro
      Convener: Ekaterina Lobacheva
      • 43
        Intro
      • 44
        Seminar: Intro
    • 1:00 PM
      Lunch
    • Section 3. Bayesian Deep Learning: Full Bayesian Inference
      Convener: Ekaterina Lobacheva
      • 45
        Full Bayesian Inference
      • 46
        Seminar: Full Bayesian Inference
    • 3:00 PM
      Break
    • Section 3. Bayesian Deep Learning: Bayesian linear regression
      Convener: Nadya Chirkova
    • 4:00 PM
      Break
    • Guest lectures
    • Section 3. Bayesian Deep Learning: Variational Inference
      Convener: Ekaterina Lobacheva
      • 47
        Variational Inference
      • 48
        Seminar: Variational Inference
    • 10:00 AM
      Break
    • Section 3. Bayesian Deep Learning: Gaussian Processes
      Convener: Nadya Chirkova
    • 11:30 AM
      Break
    • Section 3. Bayesian Deep Learning: Bayesian Neural Networks
      Convener: Nadya Chirkova
      • 49
        Bayesian Neural Networks - Introduction
      • 50
        Bayesian Neural Networks - Training
    • 1:30 PM
      Lunch
    • Section 3. Bayesian Deep Learning: Bayesian Neural Networks
      Convener: Nadya Chirkova
      • 51
        Seminar: Bayesian Neural Networks - Training
      • 52
        Bayesian Sparsification of Neural Networks
    • 3:30 PM
      Break
    • 53
      Introduction into the 2nd coopetition
      Speaker: Artem Maevskiy (National Research University Higher School of Economics (RU))
    • Guest lectures
    • Section 3. Bayesian Deep Learning: VAE
      • 54
        VAE
      • 55
        Seminar: VAE
    • Section 3. Bayesian Deep Learning: Seminar: VAE
      Convener: Alexei Struminsky (Space Research Institute)
    • 10:30 AM
      Break
    • Section 4. Generative models and networks: Generative models
      Convener: Nikita Kazeev (Yandex School of Data Analysis (RU))
    • Section 4. Generative models and networks: Seminar: Practice on basic generative models
      Convener: Nikita Kazeev (Yandex School of Data Analysis (RU))
    • 12:00 PM
      Break
    • Section 4. Generative models and networks: Introduction to distances Pt. 1
      Convener: Vladislav Belavin (Yandex School of Data Analysis (RU))
    • Section 4. Generative models and networks: Seminar: Distances Pt. 1
      Convener: Vladislav Belavin (Yandex School of Data Analysis (RU))
    • 1:30 PM
      Lunch
    • Section 4. Generative models and networks: Introduction to distances Pt. 2
      Convener: Vladislav Belavin (Yandex School of Data Analysis (RU))
    • Section 4. Generative models and networks: Seminar: Distances Pt. 2
      Convener: Vladislav Belavin (Yandex School of Data Analysis (RU))
    • 3:30 PM
      Break
    • Guest lectures
    • Section 4. Generative models and networks: Autoencoders
      Convener: Artem Ryzhikov (National Research University Higher School of Economics (RU))
      • 56
        Autoencoders
      • 57
        Seminar: Practice on AE
    • 10:00 AM
      Break
    • Section 4. Generative models and networks: GANs
      Convener: Nikita Kazeev (Yandex School of Data Analysis (RU))
      • 58
        GANs
      • 59
        Seminar: GANs
    • 11:30 AM
      Break
    • Section 4. Generative models and networks: Advanced GANs
      Convener: Nikita Kazeev (Yandex School of Data Analysis (RU))
      • 60
        Advanced GANs
      • 61
        Seminar: Advanced GANs
    • 1:00 PM
      Lunch
    • Section 4. Generative models and networks: Flow models
      Convener: Artem Ryzhikov (National Research University Higher School of Economics (RU))
      • 62
        Flow models
      • 63
        Seminar: Flow models
    • 3:00 PM
      Break
    • Section 4. Generative models and networks: Invertible Generative Models
      • 64
        Invertible Generative Models
      • 65
        Seminar: Invertible Generative Practice
    • 4:30 PM
      Break
    • Guest lectures
    • Section 5. Advanced Optimization Methods: Introduction to black-box optimization
      Convener: Maxim Borisyak (Yandex School of Data Analysis (RU))
      • 66
        Introduction to black-box optimization
      • 67
        Seminar: Introduction to black-box optimization
    • 10:00 AM
      Break
    • Section 5. Advanced Optimization Methods: Variational Optimization
      Convener: Mr Vladislav Belavin (Yandex School of Data Analysis (RU))
    • 11:30 AM
      Break
    • Guest lectures
    • Section 5. Advanced Optimization Methods: Bayesian Optimization
      Convener: Maxim Borisyak (Yandex School of Data Analysis (RU))
      • 68
        Bayesian Optimization
      • 69
        Seminar: Bayesian Optimization
    • 10:00 AM
      Break
    • Section 5. Advanced Optimization Methods: BO-GP and friends-1
      Convener: Maxim Borisyak (Yandex School of Data Analysis (RU))
      • 70
        BO-GP and friends
      • 71
        Seminar: BO-GP and friends
    • 11:30 AM
      Break
    • Section 5. Advanced Optimization Methods: BO-GP and friends-2
      Convener: Maxim Borisyak (Yandex School of Data Analysis (RU))
      • 72
        BO-GP and friends
      • 73
        Seminar: BO-GP and friends
    • 1:00 PM
      Lunch
    • Guest lectures
    • 3:30 PM
      Break
    • Socialization: Open flow discussion
    • Section X: Learning to pivot
      Convener: Maxim Borisyak (Yandex School of Data Analysis (RU))
      • 74
        Learning to pivot
      • 75
        Seminar: Learning to pivot
    • 10:00 AM
      Break
    • Section X: Deep Learning at scale
    • 12:30 PM
      Lunch
    • Section X: Interpretability
    • 2:30 PM
      Break
    • Guest lectures
      Convener: Michael Aaron Kagan (SLAC National Accelerator Laboratory (US))
    • 4:30 PM
      Break
    • 76
      Guest Lecture
      Speaker: Dr Michela Paganini (Facebook AI Research)
    • Section X: Anomaly Detection-1
      Convener: Denis Derkach (National Research University Higher School of Economics (RU))
    • 10:00 AM
      Break
    • Section X: Anomaly Detection-2
      Convener: Denis Derkach (National Research University Higher School of Economics (RU))
    • 11:30 AM
      Break
    • Guest lectures
    • 1:30 PM
      Lunch
    • 2:30 PM
      Break
    • Guest lectures
      • 4:00 PM
        Lunch
    • Guest lectures
    • 10:30 AM
      Break
    • Socialization: Student's project presentations
    • 77
      Closing words
      Speaker: Andrey Ustyuzhanin (Yandex School of Data Analysis (RU))
    • 1:30 PM
      Lunch
    • Socialization: Follow-up (VR?) party