On November 28, Huy Vo defended his PhD thesis entitled “Annotation-efficient learning for object discovery and detection". Congratulations!
The Cifre thesis of Van Huy Vo was supervised by Patrick Perez from Valeo.ai and Jean Ponce, PR[AI]RIE Scientific Director and chair holder.
Here is a short abstract of the thesis: Object detectors are important components in many intelligent systems. They are typically obtained with fully-supervised training which require costly annotated data. This thesis considers annotation-efficient alternatives to fully-supervised object detection. In particular, we discuss several approaches to Unsupervised Object Discovery, which aims to link images containing similar objects and localize these without any supervision. We also introduce a framework which combines active learning and weakly-supervised object detection that demonstrates a better detection performance / annotation cost trade-off than both fully- and weakly-supervised object detection.
COFUND “Artificial Intelligence for the Sciences” : Call for PhD projects open!
Artificial Intelligence for the Sciences (AI4theSciences) is a doctoral program run by Université PSL, within which more than 20 PhD contracts at the interfaces of artificial intelligence or big data processing are offered.
Following this call, already a 3rd cohort of students will begin their PhD, starting in autumn 2023. The deadline for applications is 31 January. More information is available here.
AI4theSciences is supported and cofunded by the Horizon 2020-Marie Skłodowska-Curie Actions-COFUND European program.
Laura Cantini wins Claude Paoletti Prize
Laura Cantini, Pr[AI]RIE chair and CNRS researcher at the Institute of Biology of the Ecole Normale Supérieure (IBENS) CNRS/ ENS/ Inserm received the Paoletti Prize for her research work.
The second laureate is Clément Charenton, CNRS researcher at the Institute of Genetics and Molecular and Cell Biology (IGBMC) CNRS/Inserm/Université de Strasbourg. The award ceremony was held on 9 November 2022.
This prize, awarded in memory of Claude Paoletti, former director of the life sciences department of the CNRS, aims to support young researchers in biology. Read more here.
Justine Cassell at the 3rd Trilateral Symposium on AI
What does it take to use AI appropriately? Experts from France, Germany and Japan met last week in Tokio at the 3rd Trilateral Symposium to discuss strategies to prepare society for the major role that AI will play in the future.
Justine Cassell, PR[AI]RIE chair, was there representing French Digital Council as PR[AI]RIE and Inria.
In the beginning of November, Justine Cassell also took part in The AI Quorum Retreat in Abu Dhabi that met to discuss the future of AI. The AI Quorum is a series of high-level gatherings designed to stimulate cutting-edge AI research with leading scientists and share an understanding of the discipline as a force for good.
Florian Yger about “Arts & AI”
Launched in October 2021 by Dauphine Numérique, Ex Machina podcasts return with 10 new episodes, during which the researchers discuss the impact of AI in the fields of art, history, health, the environment and geopolitics.
Santé, finance, éthique, régulation, médias… Comment l’intelligence artificielle transforme-t-elle en profondeur notre société ?
L’Université Paris Dauphine – PSL, en partenariat avec l’institut 3IA PRAIRIE et Les Échos, vous invite à : 3 jours de conférences-débatssur les impacts de l’IA
Avec la participation de scientifiques de l’Université Paris Dauphine – PSL, d’expertes et d’experts internationaux du domaine (médaillés Fields et du CNRS, chercheurs de Berkeley, Washington University, membres de l’OMS, de l’Académie des Sciences, du Conseil d’État, journalistes de l’AFP, des Échos…) :
5 THÉMATIQUES :
LA RÉGULATION DES
PLATEFORMES – Lundi 21 nov. (09H-12H30)
ÉTHIQUE DE L’IA – Lundi 21 nov. (14H-19H)
SANTÉ NUMÉRIQUE –
Mardi 22 nov. (09H-12H)
FINANCE ET IA –
Mardi 22 nov. (14H-18H)
L’IMPACT DE L’IA DANS
LES MÉDIAS – Mercredi 23 nov. (09H-12H30)
ET LA SOIRÉE DAUPHINE DIGITAL DAYS – Mardi 22 nov. (18H-22H)
Jean-Baptiste Masson, Principal investigator of the Decision and Bayesian Computation Laboratory (Pasteur – CNRS), AVATAR MEDICAL CSO and Co-founder, and PR[AI]RIE chair, is among the three researchers who have been awarded “Prix des Innovateurs 2022” / Innovators Award 2022 by the Île-de-France Region.
Since 2020, the Innovators’ Prize created by the Île-de-France Region, aims to encourage, support and promote research, technology transfer and innovation in the medical field.
Each year, the Innovators’ Prize rewards three researchers under 45 years of age and their teams for the quality of their work in the field of health. A total of €100,000 award aims to help them advance their research work. Read more here.
New PR[AI]RIE PhD thesis defended
On September 23, Luca Ganassali defended his doctoral thesis entitled “The graph alignment problem: fundamental limits and efficient algorithms ". Congratulations!
His supervisors were two PR[AI]RIE chairs: Laurent Massoulié and Marc Lelarge.
The thesis studies the graph alignment problem, the noisy version of the graph isomorphism problem, which aims to find a matching between the nodes of two graphs which preserves most of the edges. Focusing on the planted version where the graphs are random, we are interested in understanding the fundamental information-theoretical limits for this problem, as well as designing and analyzing algorithms that are able to recover the underlying alignment in the data. For these algorithms, we give some theoretical high probability guarantees of the regime in which they succeed or fail.
Bodies and AI, deeper links than we think: interview with Justine Cassell
For Justine Cassell, an expert in human-machine interaction, AI embodied in machines is an effective tool for developing communication between humans, through the body.
We interact with robots unconsciously in the same way that we interact with humans using our bodies. For example, we unknowingly monitor eye movements, hand gesture and smile, and draw conclusions about the trustworthiness of the person in front of us and about their engagement in the conversation. This ability to use the body to engage in getting to know the other person begins at birth and lasts throughout life. It’s exciting to realise that you do the same thing with a robot!
Dauphine Numérique, a programme whose ambition is to promote a world-class research and training ecosystem and to produce interdisciplinary work on digital transformation, invites you to participate in the Dauphine Digital Days, a series of conferences on AI.
Three days of talks and roundtables clustered around 5 main themes will take place on 21, 22 and 23 November 2022 at the University of Paris Dauphine – PSL. The themes proposed are: Digital Plateform Regulation, Justice and Fairness Issues, Digital Health, AI and Finance, and Digital and Information.
More info here. Don’t miss information on registration coming up early September!
Best Paper Award for Umut Simsekli
Congratulations to Umut Simsekli, PR[AI]RIE chair, who won Best Paper Award for the article "Generalized Sliced Probability Metrics" at the IEEE ICASSP conference held in May 2022.
The International Conference on Acoustics, Speech and Signal Processing is the flagship event of the IEEE Signal Processing Society. Read the winning article here.
Runner-up best paper award at ISBI
Benoît Sauty, a PhD student in the ARAMIS team, was awarded the Best-Paper Runner-Up Award at the ISBI Conference, for the paper entitled "Riemannian Metric Learning for Progression Modeling of Longitudinal Datasets".
ISBI is the International Symposium of Biomedical Imaging, one of the world’s largest conferences for the medical imaging community. Congratulations! Read the full paper here.
Top contributor award from Papers with Code
Clément Chadebec, a PhD student under the supervision of Stéphanie Allassonnière, received a top contributor award from Papers with Code for the python library unifying generative autoencoder implementations and focusing on reproducibility.
Pythae, is a versatile open-source Python library providing both a unified implementation and a dedicated framework allowing straightforward, reproducible and reliable use of generative autoencoder models. Check it here.
“AI for biomedical imaging” working group in “Le Monde”
In the article published in the Science & Medicine (supplement of “Le Monde”) the inter 3IA institutes working group on "IA for biomedical imaging” composed of representatives of 3IA Côte d’Azur, MIAI@Grenoble Alpes and PR[AI]RIE, proposes its view on the future of medical imaging thanks to AI.
Medicine has been a major field of experimentation for artificial intelligence from the start. Nowadays the potential of AI is proven, but the tools are not widely used in the medical environment.
While AI holds the promise of major advances in medical imaging, its use requires multidisciplinary thinking, argue researchers in applied mathematics, computer science, physics, neuroscience and medical science. Rather than fearing that doctors will be replaced by artificial intelligence, a question should be asked what kind of services we want to delegate to AI.
What are the obstacles to the wider deployment of AI
in healthcare, and where do research efforts lie? Read
more in the article.
Applications for PRAIRIE Master fellowships 2022/23 are now open!
PRAIRIE Master’s fellowships are accepting applications for four attractive programs:
The fellowships are available for M1 or M2 years and are awarded on
academic criteria. Depending on the rules of each university, the
fellowship amount can be maximum 1000€ per month.
Please find here the forms which need to be attached to your application dossier.
Colloquium « Génération IA 2030 »
Several PRAIRIE chair holders participated in the colloquium « Génération IA 2030 » which took place on 17 November in Brussels.
The colloquium was organised by the French Embassy in Belgium and AI4Belgium with the objective to support the Belgian initiatives in the field of Artificial Intelligence. The event brought together political, scientific and economic actors playing a major role in the field of artificial intelligence.
Digital technology and Artificial Intelligence are one of the priority topics for the French Presidency of the EU Council.
Award at AAAI 2022
TheAssociation for the Advancement of Artificial Intelligence, AAAI-22, has just announced the winners of several awards. The Outstanding Paper Award was given to a team featuring PR[AI]RIE Deputy Scientific Director Jamal Atif and PR[AI]RIE PhD student Virginie Do for a paper entitled “Online certification of preference-based fairness for personalized recommender systems”. Congratulations!
A groundbraking study on methane emissions, based on Kayrros data, has been published in a top-tier scientific journal “Science”, and attracted extensive media coverage and recognition from the French President Emmanuel Macron.
The study provides first statistical characterisation of methane ultra-emitters from oil and gas. It is based on Kayrros data which provides the most comprehensive measure to date of methane, a potent greenhouse gas. Methane emissions from oil and gas production and transmission represent a significant contribution to climate change. These emissions comprise sporadic releases of large amounts of methane during maintenance operations or equipment failures not accounted for in current inventory estimates. The study shows a direct relationship between ultra-emitters monitored and measured from space and smaller methane leaks detected by local sensors and aircraft surveys.
The study contributor, Kayrros co-founder and scientific director is Alexandre d’Aspremont, PR[A]IRIE chair. Congratulations!