SOSC 2019 Third International School on Open Science Cloud

16 Sept 2019, 08:00 → 20 Sept 2019, 22:00 Europe/Rome

Centro nazionale dell'INFN per la ricerca e lo sviluppo nelle tecnologie informatiche e telematiche (Bologna, CNAF)

Daniele Bonacorsi (BO), Daniele Spiga (PG), Davide Salomoni (CNAF), Livio Fano' (PG), Luca Scrucca (Università di Perugia), Mirko Mariotti (PG), Valentin Kuznetsov (Cornell University (US))

The theme of the third International School on Open Science Cloud is “Intelligent Systems”

Bulletin 1 SOSC19.pdf

Bulletin 2 SOSC19.pdf

Poster SOSC19.pdf

Monday 16 September

1 Welcome message and Introduction to the School

Speakers: Dr Gaetano Maron (INFN CNAF and Legnaro), Dr Davide Salomoni (INFN CNAF)

Introduction to SOSC2019.pdf

Lectures: Introduction

2 Introduction to Statistical Learning

Statistical learning refers to a vast set of tools and techniques for modelling and understanding complex datasets. A great deal of interest on these aspects has recently surged in the statistical and computer science communities, and statistical learning has become a very hot field in many applications, such as marketing, finance, and other business disciplines. In this short introduction we review basic ideas of supervised and unsupervised learning, classification and regression, statistical models for inference and prediction, interpretability vs flexibility, bias-variance trade-off, overfitting.

Speaker: Prof. Luca Scrucca (University of Perugia)

A_Not_So_Short_Intro_To_SL_slides2up.pdf

3 Introduction to Intelligent Systems

Intelligent systems are ubiquitous. Their growth and deployment are addressing industrial, scientific and societal challenges. A brief overview of the applications of intelligent systems and predictive models - with focus on current and future opportunities - is offered. A major focus is given on scientific challenges in medical sciences, chemistry, bioinformatics, geosciences and physics.

Speaker: Prof. Daniele Bonacorsi (University of Bologna)

SOSC19_DBonacorsi_IntelligentSystems.pdf

11:15 Coffee Break

Lectures: Introduction

4 Introduction to Intelligent Infrastructures

What do we mean by "data and compute infrastructures" and to what extent they can be considered "Intelligent"? How and why are these terms changed over the past 20 years? This lecture will introduce the main results and challenges for efficient exploitation of data-driven research.

Speaker: Dr Davide Salomoni (INFN CNAF)

Intelligent Infrastructures SOSC2019.pdf

12:45 Lunch break

Hands-on: Data Science

https://cloud-pg.github.io/SOSC2019

5 Software ecosystem for Data Science

Speakers: Prof. Daniele Bonacorsi (University of Bologna), Dr Valentin Kuznetsov (Cornell University)

HandsOnMaterials.pdf

16:30 Coffee Break

Hands-on: Mini Kaggle

https://cloud-pg.github.io/SOSC2019

6 Introduction to the mini-kaggle competition

Speakers: Dr Valentin Kuznetsov (Cornell University), Prof. Daniele Bonacorsi (University of Bologna)

Tuesday 17 September

Lectures: Machine Learning

7 Data Science: state of the art

The speaker will present the state of the art in data science, the role of (big) data in it, the skills needed to perform efficiently as a data scientist, and the key concepts around data, algorithms and techniques in the field.

Speaker: Dr Valentin Kuznetsov (Cornell University)

Kuznetsov-Tuesday.pdf

8 Machine/Deep Learning in HEP

The speaker will give an overview of the utilisation of ML/DL techniques in High Energy Physics, with and without accelerators, giving (selected) examples. The talk will present the impact of such techniques in triggering, tracking and reconstruction, end-user data analysis, computing infrastructure optimization, etc. The talk will also discuss the transition from the use of field-specific knowledge for feature engineering to the use of full high-dimensional feature space to train cutting-edge ML algorithms, and will discuss possible evolutions of the role of ML/DL in HEP.

Speaker: Prof. Daniele Bonacorsi (University of Bologna)

SOSC19_DBonacorsi_MLDLinHEP.pdf

11:00 Coffee Break

Lectures: Statistical Learning

9 Resampling Methods

Choosing a proper level of complexity for a prediction rule (model selection) or evaluating its performance (model assessment) are two fundamental steps in any supervised statistical learning application. Both steps require reliable estimates of the expected prediction error. After providing some general definitions related to the prediction error in regression and classification, attention will be focused on estimating this error by re-using the observed data. In particular, methods based on validation sets, cross-validation and bootstrap will be illustrated. Advantages and disadvantages of each technique will be discussed.

Speaker: Prof. Giuliano Galimberti (University of Bologna)

Resampling_Methods_slides2up.pdf

12:30 Lunch Break

Hands-on: How to become a ML model guru

https://cloud-pg.github.io/SOSC2019

Conveners: Prof. Daniele Bonacorsi (University of Bologna), Dr Valentin Kuznetsov (Cornell University)

16:30 Coffee Break

Hands-on: Work on the the mini-kaggle competition

https://cloud-pg.github.io/SOSC2019

Conveners: Dr Valentin Kuznetsov (Cornell University), Prof. Daniele Bonacorsi (University of Bologna)

Wednesday 18 September

Lectures: Sigillum of the Alma Mater to Professor Trevor Hastie, see https://eventi.unibo.it/sigillumhastie

12:30 Lunch Break

Hands-on: Mini kaggle intermediate wrap-up, and a new challenge

https://cloud-pg.github.io/SOSC2019

Conveners: Prof. Daniele Bonacorsi (University of Bologna), Dr Valentin Kuznetsov (Cornell University)

16:00 Coffee Break

Hands-on: Mini-kaggle + image classification (Part I)

https://cloud-pg.github.io/SOSC2019

Conveners: Dr Valentin Kuznetsov (Cornell University), Prof. Daniele Bonacorsi (University of Bologna)

10 Hands-on: Mini-kaggle + image classification (Part I)

Speaker: Dr Valentin Kuznetsov (Cornell University (US))

Neural Network Zoo

SOSC19_DBonacorsi_TPU.pdf

Visual explanation of CNN networks

Thursday 19 September

Lectures: Computing Infrastructures

11 Containers, their composition and orchestration

Speaker: Dr Davide Salomoni (INFN CNAF)

Containers SOSC2019.pdf

Hands-on: Containers

https://cloud-pg.github.io/SOSC2019

Convener: Dr Davide Salomoni (INFN CNAF)

11:00 Coffee Break

Hands-on: Hands-on: Containers

https://cloud-pg.github.io/SOSC2019

12:30 Lunch Break

Lectures: Computing Infrastructures

12 Function as a Service and data streaming analytics in cloud infrastructures.

Speaker: Dr Daniele Spiga (INFN Perugia)

FaaS-Serverless-SOSC19.pdf

Lectures: Computing Infrastructures

13 Infrastructure as Code and Template-based orchestration in hybrid Clouds.

Speaker: Mirco Tracolli (PG)

InfrastructureAsCode.pdf

16:30 Coffee Break

Hands-on: FaaS, analytics and template-based orchestration.

https://cloud-pg.github.io/SOSC2019

Convener: Diego Ciangottini (PG)

14 Hands-on: FaaS, analytics and template-based orchestration.

Speaker: Diego Ciangottini (PG)

hands-on guide

20:30 Social Dinner

Friday 20 September

Hands-on: (Mini-kaggle + image classification (Part II))

https://cloud-pg.github.io/SOSC2019

15 (Mini-kaggle + image classification (Part II))

Speaker: Dr Valentin Kuznetsov (Cornell University (US))

11:00 Coffee Break

Lectures: Wrap-up

16 Wrap-up of the school and follow-up opportunities for students

Wrap up SOSC2019.pdf

12:30 Lunch Break

Final Exam and results of the Kaggle Competition