Sociétés durables et résilientes

L'humanité est confrontée à de nombreux défis, notamment causés par le changement climatique, les guerres et les pandémies. Le progrès technologique peut être mis à profit pour le bien commun sur la voie de sociétés résilientes et durables.

Les scientifiques de l'Idiap travaillent sur l'intégration de la dimension humaine dans la conception et l'utilisation de l'intelligence artificielle. Ils apportent une expertise multidisciplinaire pour permettre l'application de solutions technologiques à une variété d'applications pertinentes pour la société dans son ensemble. Par exemple, leurs travaux portent sur la lutte contre la désinformation, le développement de méthodes d'économie d'énergie ou les mesures visant à aider la société à agir rapidement face aux risques prédits.

Domaines d'expertise

#Bioinformatics&HealthInformatics
#DataScience&SocialComputing
#HumanComputerInteraction
#Imaging&ComputerVision
#MachineLearning
#NaturalLanguageProcessing
#Robotics&AutonomousSystems
#Security&Privacy
#SignalProcessing
#Speech&AudioProcessing

Plus  

Ce programme contribue aux ODD des Nations-Unies suivants


Plus  

Personnes

Nothing to list

 

Publications choisies

Integrating daylight with general and task lighting: A longitudinal in-the-wild study in individual and open space working areas, Chantal Basurto, Michael Papinutto, Moreno Colombo, Roberto Boghetti, Kornelius Reutter, Julien Nembrini and Jérôme Kämpf, in: Solar Energy Advances, 2, 2022.

This paper makes use of AI-based surrogate models to predict the indoor lighting conditions and control optimally the blinds and electric lighting to maintain visual comfort and achieve energy savings. More than 50% of electricity for lighting were saved without impacting significantly visual comfort over the course of our longitudinal experiment.

 

Comprehensive Vulnerability Evaluation of Face Recognition Systems to Template Inversion Attacks Via 3D Face Reconstruction, H. S. Otroshi and S. Marcel, IEEE TPAMI 2023.

In this work, we propose a new method (called GaFaR) to reconstruct 3D faces from facial templates using a pretrained geometry-aware face generation network, and train a mapping from facial templates to the intermediate latent space of the face generator network. We train our mapping with a semi-supervised approach using real and synthetic face images. For real face images, we use a generative adversarial network (GAN)-based framework to learn the distribution of generator intermediate latent space. For synthetic face images, we directly learn the mapping from facial templates to the generator intermediate latent code. We demonstrated the transferability of our attack with state-of-the-art methods across other face recognition systems. We also performed practical presentation attacks on face recognition systems using the digital screen replay and printed photographs, and evaluated the vulnerability of face recognition systems to different template inversion attacks.

 

Claim-Dissector: An Interpretable Fact-Checking System with Joint Re-ranking and Veracity Prediction, Martin Fajcik, Petr Motlicek and Pavel Smrz, in: Association for Computational Linguistics, Findings of the Association for Computational Linguistics: ACL 2023:10184–10205, 2023.

This paper describes new latent variable model for fact-checking and fact-analysis, which given a claim and a set of retrieved provenances allows learning jointly: (i) what are the relevant provenances to this claim (ii) what is the veracity of this claim. We propose to disentangle the per-provenance relevance probability and its contribution to the final veracity probability in an interpretable way - the final veracity probability is proportional to a linear ensemble of per-provenance relevance probabilities. This way, it can be clearly identified the relevance of which sources contributes to what extent towards the final probability. We show that our system achieves state-of-the-art results on FEVER dataset comparable to two-stage systems typically used in traditional fact-checking pipelines, while it often uses significantly less parameters and computation.

 

Full list of related projects

Eguzki and IVECT, 2020-2023, SFOE, Kämpf

Built environment sustainability

 

SOTERIA, 2022-2024, EU, Marcel

Face recognition anti-spoofing

 

GRAIL, 2022-2025, US IARPA, Marcel

Person recognition at a distance

 

TRESPASS, 2020-2024, EU, Marcel

Biometrics security and privacy preservation

 

CRiTERIA, 2021-2024, EU, Motlicek

Comprehensive data-driven risk and threat assessment methods for the early and reliable Identification, validation and analysis of migration-related risks

 

ROXANNE, 2019-2023, EU, Motlicek

Real-time network, text and speaker analytics for combating organized crime

 

TRACY, 2023-2025, EU, Motlicek

Big-data analytics from base-stations registrations and e-evidence system