MCnet Machine Learning School

22 Jun 2020, 08:05 → 26 Jun 2020, 18:00 Europe/Stockholm

Zoom (Lund University)

We are happy to announce that the MCnet Machine Learning training event will be held online from Lund University, June 22nd-June 26th 2020.

The event provides a five day course of lectures, talks and tutorials on Machine Learning with applications from particle physics as well as industry.

This is a training event within the MCnetITN3 Innovative Training Network, primarily directed to students and early stage postdocs affiliated with MCnet. Applications from outside of MCnet are welcome and will be considered if places are available. We aim at approximately 30 students, and the event is free of charge.

Our core sessions comprise a series of introductory lectures given by Professor Mattias Ohlsson (Lund University). Our preliminary list of lecturers/speakers include:

• Historical notes - Carsten Peterson (Computational Biology, Lund)
• Introduction to machine learning - Mattias Ohlsson (Computational Biology, Lund)
• Machine learning in high energy physics - Ben Nachman (LBNL, Berkeley)
• Machine learning for image analysis - Niclas Danielsson (Axis, Lund)
• How to GAN LHC events - Anja Butter (Heidelberg),
• Accelerating HEP theory with ML models - Stefano Carrazza (Milan)
• How to implement denoising and variational autoencoders - Najmeh Abiri (IT-University Copenhagen Computer Science)
• Bayesian deep probabilistic differentiable programming: A scientific approach to AI - Michael Green (Desupervised, Copenhagen)
• Can we “machine-learn” the next standard model?  - Wolfgang Waltenberger (Vienna)
• Towards the autonomous machine learning fueled supply chain - Malte Tichy (Blue Yonder, Hamburg)
• Outlook on ML in HEP - Tilman Plehn (Heidelberg)

---

 

Organised by:

---

Monday 22 June

1 Welcome from the Organisers

Speakers: Malin Sjödahl, Stefan Prestel

talk20200622.pdf

2 Historical Perspective

Speaker: Prof. Carsten Petersson (Lund University)

Carsten_Peterson_2020_MCnet_Lund.pdf

09:30 Question Break

3 General introduction

Speaker: Mattias Ohlsson

Intro-part1.pdf

10:00 Question Break

4 General introduction (continued)

Speaker: Mattias Ohlsson

5 Introduction to Machine Learning

Speaker: Mattias Ohlsson

Intro-part2.pdf

10:35 Feedback and coffee

6 Introduction to Machine Learning (continued)

Speaker: Mattias Ohlsson

7 The MLP Architecture

Speaker: Mattias Ohlsson

Intro-part3.pdf

11:25 Question Break

8 The MLP Architecture (continued)

Speaker: Mattias Ohlsson

12:00 Free time

9 The CNN architecture

Speaker: Mattias Ohlsson

Intro-part4.pdf

14:00 Question Break

10 The CNN architecture (continued)

Speaker: Mattias Ohlsson

Tuesday 23 June

11 Welcome

Speaker: The organizers

12 Recurrent neural networks

Speaker: Mattias Ohlsson

Intro-part5.pdf

09:40 Question Break

13 Recurrent neural networks (continued)

Speaker: Mattias Ohlsson

10:15 Coffee break

Tutorials: Exercise

Conveners: Mattias Ohlsson, Najmeh Abiri

Intro-Lab.zip

12:00 Free time

Tutorials: Tutorial Session 2

Conveners: Mattias Ohlsson, Najmeh Abiri

Intro-Lab.zip

15:00 Free time

14 ML in HEP: preliminaries

Speaker: Ben Nachmann

LundMCNetSchool_1.pdf

17:30 Question Break

15 Deep learning with HEP images

Speaker: Ben Nachmann

18:05 Question break

16 Deep learning in HEP beyond images

Wednesday 24 June

17 Welcome

Speaker: The organizers

18 Machine learning for image analysis: Recap + Deep Learning for Video and Audio

Some useful links, in case zoom video sharing is not possible with your connection. We will also post hem in the chat.

Here are all the links in the correct order as presented.

YOLO v2 detector (intro):
https://www.youtube.com/watch?v=VOC3huqHrss

Semantic segmentation Cityscapes
https://www.youtube.com/watch?v=ATlcEDSPWXY

Object Detetection, traffic intersection:
https://www.youtube.com/watch?v=F-lWyJ5Trk4

​Instance segmentation (YOLACT):
https://www.youtube.com/watch?v=0pMfmo8qfpQ​

​Pose estimation on top of Mask RCNN instance segmentation:
https://www.youtube.com/watch?v=KYNDzlcQMWA​

Speaker: Niclas Danielsson

Machine Learning for Image Analysis_with_movies_as_links.pdf

09:40 Question Break

19 Machine learning for image analysis: Deep Learning in the Industry and Deployment Platforms

Some useful links, in case zoom video sharing is not possible with your connection. We will also post hem in the chat.

Here are all the links in the correct order as presented.

YOLO v2 detector (intro):
https://www.youtube.com/watch?v=VOC3huqHrss

Semantic segmentation Cityscapes
https://www.youtube.com/watch?v=ATlcEDSPWXY

Object Detetection, traffic intersection:
https://www.youtube.com/watch?v=F-lWyJ5Trk4

​Instance segmentation (YOLACT):
https://www.youtube.com/watch?v=0pMfmo8qfpQ​

​Pose estimation on top of Mask RCNN instance segmentation:
https://www.youtube.com/watch?v=KYNDzlcQMWA​

Speaker: Niclas Danielsson

10:15 Feedback and coffee

20 Machine learning for image analysis: Introduction to Tensorflow 2

Some useful links, in case zoom video sharing is not possible with your connection. We will also post hem in the chat.

Here are all the links in the correct order as presented.

YOLO v2 detector (intro):
https://www.youtube.com/watch?v=VOC3huqHrss

Semantic segmentation Cityscapes
https://www.youtube.com/watch?v=ATlcEDSPWXY

Object Detetection, traffic intersection:
https://www.youtube.com/watch?v=F-lWyJ5Trk4

​Instance segmentation (YOLACT):
https://www.youtube.com/watch?v=0pMfmo8qfpQ​

​Pose estimation on top of Mask RCNN instance segmentation:
https://www.youtube.com/watch?v=KYNDzlcQMWA​

Speaker: Niclas Danielsson

TensorFlow_2.0_MCNet_introduction.pdf

11:05 Question Break

21 Machine learning for image analysis: Preparing, training, visualizing.

Speaker: Niclas Danielsson

11:45 Feedback and coffee

12:00 Free time

Tutorial and Transfer Learning: Machine Learning for Image Analysis: Machine Learning for Image Analysis

For completeness, please find below links to the colab exercises. You'll go through these in the tutorial session.

Exercise 2: ”First Tensorflow Training Example”
https://colab.research.google.com/drive/1f8s8L4dGFP4nHz415iH5oTNh3TRCmC7h
https://colab.research.google.com/drive/12hoctT83TjimvNnwtq9O6BkNf5eKdqRi​

Exercise 3: ”Tensorboard and Learning rate Schedules”
https://colab.research.google.com/drive/107oomMnOoNL0RZYq09dUJCWe3sK026yI
https://colab.research.google.com/drive/1JyFqekeoXDNVRAZaDSx7pvLWXtA5aX-5

Exercise 4: ”Transfer Learning”
https://colab.research.google.com/drive/1U6FU7LNClWQ1NJEUDJ44NrHecRkpgzvg
https://colab.research.google.com/drive/1GUDePuCBtT-h73zMPLDOWbWAsOT2fMr9

Convener: Niclas Danielsson

Free time

22 ML in HEP: Likelihood-free methods for removing distortions

Speaker: Ben Nachmann

LundMCNetSchool_2.pdf

17:15 Break/Free time

23 ML in HEP: Generative models

Speaker: Ben Nachmann

18:45 Question break

24 ML in HEP: Uncertainty quantification and anomaly detection

Speaker: Ben Nachmann

Thursday 25 June

25 Welcome

Speaker: The organizers

26 How to GAN LHC events

Speaker: Anja Butter

GAN_LHC.pdf

09:40 Question Break

27 How to GAN LHC events (continued)

Speaker: Anja Butter

10:15 Feedback and coffee

28 Accelerating HEP theory with ML models

Speaker: Stefano Carrazza

mcnet_carrazza.pdf

11:05 Question Break

29 Accelerating HEP theory with ML models (continued)

Speaker: Stefano Carrazza

11:40 Feedback and coffee

12:00 Free time

Tutorial: Autoencoders and their applications

Convener: Najmeh Abiri

Link to github

Friday 26 June

30 Welcome

Speaker: The organizers

31 Can we "machine learn" the Next Standard Model?

Speaker: Wolfgang Waltenberger

lund.pdf

09:40 Question Break

32 Can we "machine learn" the Next Standard Model? (continued)

Speaker: Wolfgang Waltenberger

10:15 Feedback and coffee

33 Towards the autonomous machine learning fueled supply chain.

Speaker: Malte Tichy

2020-06-26_MCNET_Lund_towards_autonomous_ML_fueled_SC.pdf

11:05 Question Break

34 Towards the autonomous machine learning fueled supply chain (continued)

Speaker: Malte Tichy

11:40 Feedback and coffee

12:00 Free time

35 Bayesian deep probabilistic differentiable programming: A scientific approach to AI

Speaker: Michael Green

Link to presentation

14:00 Question Break

36 Bayesian deep probabilistic differentiable programming: A scientific approach to AI (continued)

Speaker: Michael Green

14:35 Feedback and coffee

37 Outlook on ML in HEP

Speaker: Tilman Plehn

mcnet_20.pdf

38 Farewell!

Speaker: The organizers

talk20200626.pdf