Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
Passive acoustic monitoring reveals feeding attempts at close range from soaking demersal longlines by two killer whale ecotypes
No full textInternational audienceOdontocetes depredating fish caught on longlines is a serious socio-economic and conservation issue. A good understanding of the underwater depredation behavior by odontocetes is therefore required. Historically, depredation on demersal longlines has always been assumed to occur during the hauling phase. In this study, we have focused on the depredation behavior of two ecotypes of killer whales, Orcinus orca, (Crozet and Type D) from demersal longlines around the Crozet Archipelago (Southern Indian Ocean) using passive acoustic monitoring. We assessed 74 hr of killer whale acoustic presence out of 1,233 hr of recordings. Data were obtained from 29 hydrophone deployments from five fishing vessels between February and March 2018. We monitored killer whale buzzing activity (i.e., echolocation signals) as a proxy for feeding attempts around soaking longlines. These recordings revealed that the two ecotypes were feeding at close range from soaking longlines, even when fishing vessels were not present. Our results suggest that both killer whale ecotypes are likely to depredate soaking longlines, which would imply an underestimation of their depredation rates. The implication of underestimating depredation rates is inaccurate accounting for fish mortality in fisheries' stock assessments
Autotelic Agents with Intrinsically Motivated Goal-Conditioned Reinforcement Learning: A Short Survey
No full textInternational audienceBuilding autonomous machines that can explore open-ended environments, discover possible interactions and build repertoires of skills is a general objective of artificial intelligence. Developmental approaches argue that this can only be achieved by autotelic agents: intrinsically motivated learning agents that can learn to represent, generate, select and solve their own problems. In recent years, the convergence of developmental approaches with deep reinforcement learning (rl) methods has been leading to the emergence of a new field: developmental reinforcement learning. Developmental rl is concerned with the use of deep rl algorithms to tackle a developmental problem-the intrinsically motivated acquisition of open-ended repertoires of skills. The self-generation of goals requires the learning of compact goal encodings as well as their associated goal-achievement functions. This raises new challenges compared to standard rl algorithms originally designed to tackle pre-defined sets of goals using external reward signals. The present paper introduces developmental rl and proposes a computational framework based on goal-conditioned rl to tackle the intrinsically motivated skills acquisition problem. It proceeds to present a typology of the various goal representations used in the literature, before reviewing existing methods to learn to represent and prioritize goals in autonomous systems. We finally close the paper by discussing some open challenges in the quest of intrinsically motivated skills acquisition
TousEnsemble, un site internet interactif pour les parties prenantes de l'inclusion scolaire des élèves avec Troubles du Spectre de l'Autisme : une conception itérative incluant des tests utilisateurs
No full textInternational audienc
Évaluation de l'expérience utilisateur d'un site web interactif de soutien à la collaboration famille - professionnels et interprofessionnelle pour la scolarisation des collégiens avec TSA
No full textInternational audienc
AutoExpe.jl: package Julia permettant d'automatiser des tâches répétitives d'expériences numériques et la génération de leurs tables de résultats
No full textA Framework for the Design of Secure and Efficient Proofs of Retrievability
No full textInternational audienceProofs of Retrievability (PoR) protocols ensure that a client can fully retrieve a large outsourced file from an untrusted server. Good PoRs should have low communication complexity, small storage overhead and clear security guarantees with tight security bounds. The focus of this work is to design good PoR schemes with simple security proofs. To this end, we propose a framework for the design of secure and efficient PoR schemes that is based on Locally Correctable Codes, and whose security is phrased in the Constructive Cryptography model by Maurer. We give a first instantiation of our framework using the high rate lifted codes introduced by Guo et al. This yields an infinite family of good PoRs. We assert their security by solving a finite geometry problem, giving an explicit formula for the probability of an adversary to fool the client. Moreover, we show that the security of a PoR of Lavauzelle and Levy-dit-Vehel was overestimated and propose new secure parameters for it. Finally, using the local correctability properties of Tanner codes, we get another instantiation of our framework and derive an analogous formula for the success probability of the audit
New Feeding Network Topology for Patch Arrays Dedicated for Spatial Intersatellite Communication
No full textInternational audienc
Combining radiometric and geometric information in spectral energy for improving water column estimation in shallow waters
No full textInternational audienc
Wheel Slip Balance Based Anti-Slip Regulation on Dissymmetric road Grip
No full textInternational audienc
