CDD, Télécom Paris, 19 Place Marguerite Perey, 91120 Palaiseau.| Entreprise/Organisme : | Télécom Paris | | Niveau d'études : | Doctorat | | Date de début : | Printemps 2025 | | Durée du contrat : | 18 ou 36 mois | | Secteur d'activité : | Intelligence Artificielle | | Description : | Post-Doc in Machine Learning (Multiple Fairness in Recommending Systems)
The group dedicated to Research in Machine Learning, Statistics & Signal Processing (the research group S2A) in Télécom Paris is recruiting a postdoc in Machine Learning (18 months contact, extendable to 36 months). The post-doc recruited will take part in an interdisciplinary collaborative research project involving the SES (Economics and Social Sciences) department of Télécom Paris and the Caisse des Dépôts et Consignations, a leading French public financial institution.
Research assignment
Research activities will focus on fairness issues for recommendation engines designed by means of machine-learning methods. With the explosion of digitized content available online, recommender systems have become an essential technology and a key element in the development of new services. In a commercial context, the algorithmic principles at work (e.g. collaborative filtering, user/content-based methods, hybrid approaches) in their operation are most often aimed exclusively at maximizing user satisfaction and increasing the platform's level of use. In the context of a public service, many other criteria and objectives must be integrated to ensure a fair service from the point of view of both users and suppliers (multi-sided fairness). It is precisely the subject of this collaborative project to propose and analyze (theoretically and empirically) methods for achieving acceptable trade-offs between the relevance of recommendations and bias mitigation. In addition to producing methodological research, the post-doc's mission will also include applied work on the current version of a deployed recommendation system, aimed at quantifying the presence of different types of bias resulting from its operation.
Keywords: public service recommender system, fair and explainable AI, bias mitigation, multi-sided fairness
Supervision: the recruit will work under the supervision of
Sephan Clémençon (https://perso.telecom-paristech.fr/clemenco/)
Winston Maxwell (https://www.telecom-paris.fr/winston-maxwell).
Charlotte Laclau (https://laclauc.github.io/)
Skills
Education : PhD in Computer Science or in Applied Maths
A short international postdoctoral experience is welcome but not mandatory
English: fluent
Expertise in Python programming, familiarity with database queries
Capacity to work in a team and develop good relationships with colleagues in other disciplines
Excellent writing and pedagogical skills
Knowledge and experience required
Research publications in Machine Learning (e.g. in Neurips, ICML, AISTATS, …)
Knowledge of how recommending systems work
Taste for AI applications and interest in its societal aspects
Additional information
The position does not involve teaching. However, on a voluntary basis, the postdoc recruited may take part in machine-learning courses (undergraduate/master level) coordinated by the supervisory team.
The position
18 months position (extendable to 36 months)
Télécom Paris, 9 place Marguerite Perey - 91120 Palaiseau - France
Application
Applicants should submit a single PDF file that includes:
motivation letter
curriculum vitae
one or two major publications
contact information for one or two references
Important dates
First-Quarter 2025: interviews with candidates (by visio-conference eventually)
Spring 2025: beginning
Contact for information/application
Stephan Clémençon stephan.clemencon@telecom-paris.fr
Charlotte Laclau charlotte.laclau@telecom-paris.fr
Winston Maxwell winston.maxwell@telecom-paris.fr
Related Websites
https://s2a.telecom-paris.fr/
www.telecom-paris.fr/ai-ethics | | En savoir plus : | https://s2a.telecom-paris.fr/
Post-Doc in Machine Learning (Multiple Fairness in Recommending Systems).pdf | | Contact : | stephan.clemencon@telecom-paris.fr |
|
