Advanced Computing and Machine Learning

25 May 2020, 08:55 → 3 Jun 2020, 18:00 Europe/Zurich

Virtual (Microsoft Teams) (Instituto Galego de Física de Altas Enerxías (IGFAE))

Andrey Ustyuzhanin (Yandex School of Data Analysis (RU)), Gonzalo Diaz Lopez (IGFAE - USC), Jose Angel Hernando Morata (Universidade de Santiago de Compostela (ES)), Josh Renner (Univ. of Valencia and CSIC (ES)), Xabier Cid Vidal (Instituto Galego de Física de Altas Enerxías)

Artificial Neural Networks (ANNs) are computational models able to learn from previously provided datasets, constituting a different paradigm than classical computation algorithms in which tasks are run sequentially. This ability has allowed ANNs to produce accurate solutions to complex problems, such as image or sound recognition, that significantly outperform solutions based on classical computing. Their development has been aided by rapid improvements in hardware capabilities over the last two decades and is nowadays an active field of study whose applications extend far beyond academic research. 
 
In this course, organized by the Instituto Galego de Física de Altas Enerxías (IGFAE), basic methods in machine learning and ANNs will be introduced at a conceptual and practical level, as well as some interesting applications in Particle Physics and industry. The course will take place virtually through Microsoft Teams in the mornings of 25th, 26th, 27th of May and 1st, 2nd and 3rd of June. It is aimed at researchers, teachers, CIT professionals and students.
 
Overview:
- Introduction to Neural Networks (Days 1, 2, 3)
- Applications in Particle Physics (Days 4 and 5)
- Applications in industry (Day 6)
 
Important requirements:
- Basic knowledge of statistics, Python and Jupyter Notebooks

Connection details:
- We will have live events in Microsoft Teams (no registration or account needed) and we will also create a team for you to join the hands-on sessions. This will be announced in due time.

To join the course every day at 9:00, click in the following links:

Day 1 [25/05/20]

Day 2 [26/05/20]

Day 3 [27/05/20]

 

Please, bear in mind we will be using the interactive rooms the second week of the course. You should be part of the specific Team created for this. If you've got problems joining, please let us know.

 

 

Connection details:
- Hands-on session link:

https://colab.research.google.com/github/HSE-LAMBDA/MLatUSC-2020

NEXT activity (Day 4):

• Slides: https://github.com/jerenner/uscnncourse/blob/master/CNNs_and_NEXT.pdf
• Code: https://colab.research.google.com/github/jerenner/uscnncourse/blob/master/next/NEXT_classification.ipynb
• Background information on the classification problem: https://jerenner.github.io/next-dnn-topology/

 

Monday, 25 May

09:00 → 12:00 Introduction into ML and linear models

AI Magic
AI vs ML
ML and Physics
Intro into python
Intro into numpy, pandas, matplotlib
Linear regression
Logistic regression

day1-linear-models.pdf

day1-MI-intro.pdf

day1-MI-intro.pptx

Tuesday, 26 May

09:00 → 12:00 Decision Trees

Decision Trees
Bootstrap, Bagging and Random Forest
Naive boosting for regression
Adaptive boosting
Gradient boosting
Stacked generalization
Bias-variance trade-off

day2-decision-trees-ensembles.pdf

day2-particle_id.pdf

day2-particle_id.pptx

Wednesday, 27 May

09:00 → 12:00 Introduction into Neural Networks

day3_deep-learning-101.pdf

Monday, 1 June

09:00 → 12:00 IGFAE 1: Convolutional NNs with NEXT

Tuesday, 2 June

09:00 → 12:00 IGFAE 2: LHCb and CALIFA

09:00 ML for event classification at LHCb

Speaker: Miguel Ramos Pernas (Universidade de Santiago de Compostela (ES))

Github LHCb ML

presentation.pdf

10:20 Break

10:40 NN in R3B/Califa

Speaker: Mr Gabriel García Jímenez (IGFAE)

NN in CALIFA

Wednesday, 3 June

09:00 → 12:00 Private companies

09:00 Triple Alpha

Speaker: Dr Alvaro Dosil Suarez (Triple Alpha)

TripleAlpha-CursoNN.pdf

10:20 Coffee break

10:40 Artabro Tech

Speaker: Miguel Vidal Marono (Artabro Tech)

Artabro_NNWorkshop_IGFAE.pdf