2-year post-doc position for the HUMANIA ANR project, aiming to further the democratization of AI. Full time job located at Université Paris-Saclay; telecommuting possible. PhD in artificial intelligence, computer vision, machine learning, or related discipline, required.
Application deadline: June 15, 2022
Date of taking office: September, 2022
Email CV and letter of motivation to Isabelle Guyon.
The aim of this post doc will be to further the democratization of AI in the framework of the HUMANIA project. While the recent successes of AI have raised both hope and fright, there is a growing concern that the societal and economic benefits of AI might increasingly be in the power and under control of a few large corporations or institutions and/or technically savvy individuals, rather than made available to the vast majority of those who could benefit from them. We are therefore committed to advancing the democratization of AI along several axes of research and engage the students with us to make “AI for All” become a reality. Our primary focus is data-driven AI a.k.a. Machine Learning. We want to progress at all levels of the AI food chain, and iterate in a virtuous cycle the steps: data collection, solution development, and application. Our objective is to leverage widely available resources, such as inexpensive laptops, cellular phones, and Internet access, to facilitate access to data-driven AI for all. However, distributing broadly such automated data-driven AI tools may potentially increase the risk of deploying faulty or harmful AI solutions, if users do not employ quality data, or if such AI solutions are not robust against adversarial attacks or make unexplainable decisions. Thus, one of the focus of this research will be to further Automated Machine Learning (AutoML) with emphasis on frugal, robust and interpretable algorithms (possibly exploiting causal modeling), which are bias-aware, fairness-aware, and privacy-aware.
Given the commitment of our team, tightly intertwined with ChaLearn, to crowdsource data-science problems via the organization of challenges, the post-doc will have the opportunity to engage in challenge organization and analyses, and contribute to the open-source project Codalab competitions. Current competition tracks include: AutoDL, meta-learning, LAP, and L2RPN. Our team is both community lead of the open-source project and provides the main public instance, which has over 50,000 users and is currently hosting on average 50 competitions per month. The recruited candidate will have the opportunity to harvest challenge results and, after separating engineering tricks from principled advances, develop and make publicly available robust and interpretable AutoML algorithms, leading to open-source software and publications.
The candidate will join a dynamic research team with diverse interests, including meta-learning, few-shot-learning, continual learning, bias in data, privacy, fairness, causality, and learning to optimize. The group leader Isabelle Guyon is Professor of Artificial Intelligence at Université Paris-Saclay, Paris, France, and president of ChaLearn, USA, a non-for-profit corporation she founded, dedicated to organizing scientific competitions. She is co-recipient of the Frontiers in Research BBVA award with Vladimir Vapnik and Bernhard Schoelkopf for the invention of SVMs, a textbook machine learning method. She also co-authored an authoritative review on feature selection (with over 18000 citations) and co-invented the Siamese neural network architecture, widely used in computer vision.
University Paris-Saclay is one of the best French and European universities, by the quality of its training and its research, through the visibility and international recognition of its 275 research laboratories and their teams, as well as by the attention given, on a daily basis and by all its staff, to the supervision, support, and development of its 48,000 students. University Paris-Saclay is made up of 10 university components, 4 Schools of Engineering (AgroParisTech, CentraleSupélec, Institut d'Optique Graduate School, ENS Paris-Saclay), a prestigious institute of mathematics (Institut des Hautes Etudes Scientifiques) and relies on 6 of the most powerful French research organizations (CEA, CNRS, INRAE, Inria, Inserm and Onera). It is associated with two universities (University of Versailles Saint-Quentin-en-Yvelines and University of Évry), which will merge in the years to come, and whose campuses are near the Saclay plateau. Its students, teacher-researchers, administrative and technical staff, and partners evolve in a privileged environment, a few kilometers from Paris, featuring all scientific domains, cutting-edge technologies, academic excellence, agriculture , historical heritage, and a dynamic industrial ecosystem. Thus University of Paris-Saclay is a leading establishment located in a vast territory where it is good to study, live and work. Website: https://www.universite-paris-saclay.fr.
Handicap-friendly establishment committed to mixity and diversity.