Machine Learning – AI Engineering- AI for the Good

Wednesday 24 June 2020 – 100% online!

Download the summary report

Registration is free but mandatory before the event  (link to the registration). Once registered, you will receive by mail a link to the live session on the pdf ticket. For technical reason, only the first 200 registrations will have access to the online session.

Quick links to sessions:

Messages from VIPs   (afternoon – 12:15-13.15)

Live plenary session   (afternoon – 13.00-17.30)

Short scientific presentations   (morning-12.15)

PROGRAM

Morning free watching of short scientific presentations

Contributions from:

• • industrial partners… on how they relate to the Summer School’s topics.
  • academia and research… on scientific topics.
  • policy partners… on the French-German cooperation.

Full list of scientific videos from academic researchers & industrial entities:

• • Mounir Atiq, Genetic algorithm for budget learning on random forests using genome reduction techniques, Tarkett & Centre Borelli, ENS Paris-Saclay.
  • Stephen Becher, Establishing privacy preferences – enabling sustainable personal research data,  University of Passau.
  • Alejandro de la Concha,  Offline detection of change-points in the mean for stationary graph signals, Centre Borelli, ENS Paris-Saclay.
  • Antoine de Mathelin,  Transfer Learning for Design Space Exploration. Application to industrial product design, Michelin & Centre Borelli, ENS Paris-Saclay.
  • Nestor Demeure,  AI augmented linear solvers: using machine learning to predict the convergence profile of a linear solver, CEA & Centre Borelli, ENS Paris-Saclay.
  • Mathilde Fekom,  Sequential control of network processes, Centre Borelli, ENS Paris-Saclay.
  • Kanishka Gosh Dastidar, Nag: Neural Aggregate Generation Framework, University of Passau.
  • Sothea Has, A clusterwise supervised learning procedure based on aggregation of distances. Applications on modeling energy data, LPSM Université de Paris.
  • Pierre Humbert, Low rank activations for tensor-based convolutional sparse coding, Centre Borelli, ENS Paris-Saclay
  • Dietmar Jakob, Impact of Voice User Interfaces on Elderly Residential Environment, University of Passau.
  • Argyris Kalogeratos, Winning the competition: enhancing counter-contagion in behavioral epidemic processes, Centre Borelli, ENS Paris-Saclay.
  • Paul Lachat, Detection and prevention of inference attacks linked to sensor data in Edge Computing Environment, INSA, France & University of Passau.
  • Alfred Laugros,  Evaluating Robustness of Neural Networks to Common Corruptions​, Atos.
  • Batiste Le Bars, Change-point detection in a time-varying Ising model, SigFox & Centre Borelli, ENS Paris-Saclay.
  • Anthea Mérida, Model pre-selection method using Neural Decision Trees, Centre Borelli, ENS Paris-Saclay.
  • Baptiste Metge, Emeric Dynomant, Miguel Garcia, Jonathan Grienay,  Covid-19 research mining, Atos.
  • Ludovic Minvielle, Step detection with convolutional models, Tarkett & Centre Borelli, ENS Paris-Saclay.
  • Eric Monchalin, Reinvent our future with the magic of AI, Atos.
  • Matthieu Ospici, Domain adaptation for person re-identification, Atos.
  • Loic Pauletto, Searching for optimal deep neural architecture, Atos.
  • Marc Platini,  Root cause analysis based on deep learning for HPC predictive maintenance, Atos.
  • Guillaume Richard, Unsupervised multi-source domain adaptation for regression, EDF & Centre Borelli, ENS Paris-Saclay.
  • Théo Saillant, Weighting method for online estimators applied to HPC job power consumption estimation from submission data, CEA & Centre Borelli, ENS Paris-Saclay.
  • Brian Tervil, A beta version of ONADAP, a visualization and decision support platform for hospital facilities in times of sanitary crisis,  Centre Borelli, ENS Paris-Saclay.
  • Charles Truong, Recent advances in change point detection, Centre Borelli, ENS Paris-Saclay
  • Sebastian Wilhelm, Activity Monitoring reusing Home Infrastructure Data, University of Passau.

Afternoon live plenary session

12:15-13.15  Messages from VIPs 

• • Juha Heikkilä, Head of Unit Robotics and Artificial Intelligence, European Commission.
  • Juliette Jarry, Vice-President of Région Auvergne Rhone Alpes in charge of Digitalisation.
  • Lionel Grotto, Directeur Général de Choose Paris Region, Région Ile de France.
  • Marco Alexander Breit, Head of Department Artificial Intelligence at the Bundesministerium für Wirtschaft und Energie. 
  • Dr. Michael Jaspers, Scientific Advisor of the German Embassy in Paris.
  • Prof. Hervé Martin, Scientific Advisor of the French Consulate in Bavaria.
  • Prof. Philippe Gréciano, Vice-President of the Deutsch-Französische Hochschule DFH-UFA.
  • Prof. Pascal Auscher, Director of the National Institute of Mathematics and their Interactions, CNRS.
  • Prof. Pierre-Paul Zalio, President of Ecole Normale Supérieure Paris-Saclay.
  • Prof. Ulrich Bartosch, President of Universität Passau.
  • Laurent Prével, Director of Ecole Nationale Supérieure d’Informatique pour l’Industrie et l’Entreprise.
  • Philippe Mareine, Senior Executive Vice President, Head of Group Human Resources, CSR, Chief Digital & Transformation Officer.
  • Dr. Natascha Eckert, Head of SIEMENS University Relations.
  • Dr. Eric Monchalin, Vice-President, Head of Machine Intelligence at Atos. 
    Show verbatim

    Accelerate the European future with the 4^th Digital French-German Summer School with Industry 2020.The Artificial Intelligence (AI) is spreading everywhere, from mobile apps becoming smarter and smarter, through objects with distributed AI, to data analytics that go far beyond predictive capabilities. Every company in every industry is under pressure to figure out how it will disrupt its business.While AI is widely adopted in Business to Consumer businesses driven by the investments of the GAFAs that are leading the research in this area, the digital revolution in the Business to Business world is starting. It offers a unique opportunity for Europe, and especially for France and Germany to remain in the lead of industrial domains, relying on their strengths in science and engineering. However, the challenges to be addressed to make them leaders in the Industry 4.0 to bring European industry to the forefront are much easier to get solved with a concerted effort.Digital twin is crucial for Industry 4.0 as it is a unique opportunity to shorten the long iterative cycles of industrial products or processes. But, particularly for the initial design phases of a product, device or process, insufficient specific data is available to train such digital models. Transfer Learning is definitely a key field to jump over this issue, offering the ability to start from the data from a similar case.The business and technological potential of AI are so huge that the prospect of machines that are self-operating and have data analysis and context-specific decision-making capacity beyond the level of human performance must be taken seriously, especially regarding its consequences, benefits and risks. The citizens’ well-being has to remain the ultimate rationale for deploying AI technologies.As a leading international IT services company, Atos is pleased to actively support this revolution with actions like, being a founding member of the “IA for Industry” Research Chair leaded by the École Normale Supérieure Paris-Saclay, or promoting the yearly French-German AI summer school since its early days. Both are mixing the strong French and German AI research communities and industrial leadership of the two countries to lead to a brilliant European future.In the difficult actual context, it is great that the 2020’s edition of the French-German AI summer school is reinventing itself and I’m excited by the agenda. It’s a valuable opportunity to move one step ahead to support the European ambitions.Éric Monchalin, Vice-President, Head of Machine Intelligence at Atos

13.00-13.15  Register & join the virtual plenary session

13.15-13.50  Official opening & group photo

• • Welcome by the organizers – Prof. Mathilde Mougeot, ENS Paris Saclay & ENSIIE, Prof. Harald Kosch & Axelle Cheney, both Passau University, Dr. Ullli Waltinger & Benno Blumoser, both Siemens AI Lab.
  • Official opening – Dr. Bernd Forster/ Michael HIinterdobler, Bayerische Staatskanzlei, Germany and by Hubert Tardieu, ATOS CEO Adviser, France,
    Show verbatim

    After Chiemsee in 2017, Cachan in 2018, Passau in 2019 we are this year fully Virtual. I do not forget that we should have been in Clermont Ferrand where Michelin has invited us for 2020. I am sure we shall be able to organize a future edition in Clermont Ferrand. AI for Industry has become a real essential topic well recognized  by the European Commission in the Public Private Partnership in preparation and we can claim to be among the first to have identified it. The chair created three years ago around ENS Paris Saclay and initially regrouping : Atos, CEA joined by Bertin has been augmented in 2018 by Michelin and SNCF followed last year by Banque de France.Inaugurating the second Summer School in 2018 I was saying: “Anticipating 2020 we are convinced that Industrial Data Platform will be common place in B-to-B in the same way platforms are dominating the B-to-C world today (in 2018)”Not quite yet on Business Data. However the importance of business data has been fully recognized in the European Strategy for Data announced on February 19, 2020 which is strongly promoting data space in a number of domains as manufacturing, energy, farming, finance , transportation, health with the objective that more industrial data be available for training machine learning.As well the GAIA-X project presented earlier by Marco Alexander Breit is a great example of franco-german cooperation on cloud which will facilitate interoperability between cloud and data sharing.On AI the initial program defined in 2017 proves to be  the right one with a specific focus on measurable AI for business as a natural next step after Analytics. The great accent put on transfer learning was also the right choice even if it proves more difficult to implement as compared to what we thought in 2017.The progress on measurable AI have been superbly illustrated by the Ile de France AI Challenge which took place at the end of 2019 with the Jury on February 26. Based upon real challenge and industrial data provided by Michelin and SNCF we were able to compare the results provided by 15 start-up’s while protecting the industrial data and executing algorithms proposed by start-up’s on a neutral machine provided by Atos. Interestingly the same company won the two challenges as different as identification of wear indicators on a tire and the forecast of traffic in subways based on past traffic.
    We did not fully anticipate the importance of ethics in AI. We have seen many attempts to give good recommendations, but they were more oriented to the consumer market. In industrial markets, we are still waiting for the proper guidance and initiative as AI for Good or Responsible AI are currently studied. As well the importance of edge computing begins to be fully realized with distribution of intelligence between edge and the center.
    I am very pleased that Siemens has accepted to organize the 2021 5^thedition in Germany and I hope that we shall be able to organize next year an AI Challenge in Germany on the same scheme as the one organized in France in 2020.
    The success of this edition with more than 200 participants collected in one week is a great encouragement to the team in charge.

13.50 – 14.50 Session 1 – Machine Learning (chaired by ENS Paris-Saclay)

• • • Live presentation by Dr. Francois Deheeger, Michelin Lead Data Scientist, France. Hybrid model in product design. slides.
    • Live presentation by Dr. Fikri Hafid, Réseau de Transport d’Electricité Head of R&D Studies, France. Machine learning use cases at a transmission system operator. slides.
    • Live presentation by Prof. Michael Granitzer, Passau University Professor for Data Science, Germany. Machine Learning Engineerging and Sustainability. slides.
    • Q&A on the 3 presentations and panel discussion.

14.50-15.00  Active Break (Stretching)

15.00-16.15  Session 2 – AI for the Good (chaired by Siemens)

• • Live presentation by Benno Blumoser, Siemens AI Lab Munich, Germany.  AI for the Good – a Corporate View.
  • Live presentation by Dr. Brian Tervil, CNRS research scientist, Borelli center, ENS Paris-Saclay, France. ONADAP – a visualization and decision support tool for dynamic human and material ressources allocation within hospital facilities in times of sanitary crisis. slides.
  • Live presentation by Valentin KRUSPEL, Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ),Germany. Data Powered Positive Deviance and Beyond – Using AI for Sustainable International Development.
  • Live presentation by Julia Gottfriedsen, German Aero-Space Center (DLR) Environmental Data Science Researcher, Germany. Using Data Analytics and Machine Learning to Control Wildfire.
  • Q&A on the 4 presentations and panel discussion.

16.15-16.20  Passive Break (get yourself a coffee before the final part)

16.20 – 17.20 Session 3 – AI Engineering (chaired by Passau University)

• • Live presentation by Prof. Gianluca Bontempi, Université Libre de Bruxelles Professor and Co-Head at Machine Learning Group, Belgium. Causality and big data analytics: risks, challenges and solutions. slides and documents.
  • Live presentation by Matthias Laporte, Inspector at Banque de France, France. Self organizing maps for anomaly detection in an operational context. slides.
  •  Live presentation by Tobias Bürger, BMW Lead Big Data and AI Platform, Germany. Machine Learning and AI in Practice – Insights into use cases from the BMW Group.
  • Q&A on the 3 presentations and panel discussion.

17.20 – 17.30  Wrap-up and Closing

Organizers

ENS Paris-Saclay, France, Prof. M. Mougeot (& ENSIIE) , Dr. A. Kalogeratos, Dr. C. Truong,
University of Passau, Germany: A. Cheney, Prof. H. Kosch,
Siemens AI Lab, Germany: Dr. U. Walltinger, Dr. B. Blumoser.

Poster in pdf

#DFGSS2020