Data chair with Givaudan
"In flavours and fragrances domains, massive data are complex and are pushing technical limits in many areas, specifically in decision-making and optimization.
The objective of this collaborative work between Givaudan and CentraleSupélec is to face the challenges “Data-to-Knowledge” and “Data-to-Decision” in an integrated way through cutting edge research.
More precisely, one aims at developing breakthrough techniques for dealing with high-dimensional and multi-scale data. Givaudan has and creates, in its activities domain, massive data of different natures. It is essential to exploit these data to improve the understanding or to provide innovative and challenging techniques in the perfumery.
Mainstream statistical models and decision-making algorithms are challenged by such heterogeneous, multi-scale, complex, incomplete and/or uncertain data. In order to generate knowledge, build models and make informed decisions, statistical validity, robustness, computational tractability and causal modelling are mandatory.
This is the scientific purpose of this Chair: developing advanced Machine Learning techniques adapted to such data."
Frédéric Pascal, professor of L2S laboratory and chairholder
AI chair with Lusis
"Lusis is the publisher of TANGO, a high-performance transactional platform. performance for payments systems and corporate finance. deal.
Based on this platform Lusis realizes systems of full payments, including fraud detection, as well as extremely rich and complex front to back trading platforms on which more than US$5 billion is now being processed per day, half of which are spread over Forex and Indices and half over the Raw materials.
Lusis and CentraleSupélec have created a research chair to strengthen their collaboration in Artificial Intelligence in the banking domain.
"The Bridgeable chair is one of 16 chairs selected by the ANR. Bridgeable, for BRIDinG thE gAp Between iterative proximaL methods and nEural networks, deals with the links existing between neural networks and certain advanced optimization concepts. It is funded to the tune of 1.3 million euros distributed between the ANR (€ 500,000) and the three partners of the chair: Schneider Electric, GE Healthcare and IFPEN (€ 800,000).
This chair aims to address two main issues: the explainability and reliability of AI based on neural networks. The application ambition is to lead to a new generation of techniques to meet the challenges arising in three fields of application: 3D medical imaging, data analysis in the field of energy and the environment and multivariate non-linear modeling of electrical systems."
Jean-Christophe Pesquet, director of CVN laboratory and chairholder
Chair with Randstad
This unprecedented academic and scientific partnership is part of a resolutely disruptive and ethical approach to the recruitment process. As in many other fields and sectors of activity, artificial intelligence, its advances and possibilities, as well as the mass and wealth of data relating to employment available, open up many prospects for improving recruitment and human resources, whether from the recruiter's point of view or that of the candidate.
A more complete and precise representation of the candidate for the job from diverse and heterogeneous data (CV, historical data, navigation data, profile on professional social networks, data from video interviews, conversations with consultants, chatbots ...). This multitude of data means building new methods of data collection and annotation to better understand the candidates and their potential.
The treatment of new recruitment tools such as audio-video interviews, through analyzing the candidate's emotions and personality. These elements can constitute a significant added value in advising candidates, at a time when recruiters' attention is focused on behavioral skills (soft skills).
The design and optimization of algorithms in the recruitment process (matching algorithms between a candidate and an offer, recommendation algorithm or techniques for finding candidates or offers) with constant attention paid to the concepts of transparency, trust and interpretability of algorithms. One of the additional challenges is that of algorithmic non-discrimination. The idea is to explore the technological and human means making it possible to guarantee that a candidate is evaluated only based on his skills and his interpersonal skills, without a subjective bias of pre-selection. intervenes in the process.
Céline Hudelot, professor of MICS laboratory
AI-RACLES chair with AP-HP
The Chair project is devoted to the exploration of frailty related to aging. The research program is structured into 3 mains axes directly linked to the massive amount data collected by the AP-HP and gathered in the EDS (Entrepôt des Données de Santé):
• Clustering from longitudinal and heterogeneous data: patient representation and phenotyping,
• Prediction and variable selection from longitudinal and heterogeneous data
• Evaluation and promotion of standardized and interoperable databases (OMOP model)
Chair with Transvalor
This novel chair is motivated by the desire to explore the possibilities of AI applied to numerical simulation. Transvalor and CentraleSupélec are thus committed to developing methods and models built from data to accelerate numerical simulations and increase the accuracy of results.
The techniques cover disciplinary fields such as numerical analysis, statistics, and databases, using machine learning methods, finite element methods, and models called 'informed by physics.'
Agile and collaborative innovation is key to supporting industries in their deep transformations and addressing future challenges, namely optimizing industrial processes, energy efficiency, production integrating circular economy, and reuse of materials.