Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
Extending the proximal point algorithm beyond convexity
No full textInternational audienceIntroduced in the 1970's by Martinet for minimizing convex functions and extended shortly afterwards by Rockafellar towards monotone inclusion problems, the proximal point algorithm turned out to be a viable computational method for solving various classes of (structured) optimization problems even beyond the convex framework. In this talk we discuss some extensions of proximal point type algorithms beyond convexity. First we propose a relaxed-inertial proximal point type algorithm for solving optimization problems consisting in minimizing strongly quasiconvex functions whose variables lie in finitely dimensional linear subspaces, that can be extended to equilibrium functions involving such functions. Then we briefly discuss another generalized convexity notion for functions we called prox-convexity for which the proximity operator is single-valued and firmly nonexpansive, and see that the standard proximal point algorithm and Malitsky’s Golden Ratio Algorithm (originally proposed for solving convex mixed variational inequalities) remain convergent when the involved functions are taken prox-convex, too
How to prepare engineering students to meet the sustainability challenges?
No full textWebinarInternational audienceThere is an increasing awareness of the society, emphasised by the current civilizational crisis all over in Europe, about the urgent need to create a more sustainable world. For developing a sustainable future, the central and determinant role of future engineers is undeniable. However, with a continually growing demand of engineering workforce, industrial and business organisations have more and more difficulties to recruit adequately trained and ready to work graduate engineers. How to attract young talents for engineering? How to prepare this new generation of holistic engineers? What skills and competences they will need to meet the sustainability challenges? The purpose of this event is to offer an overview of recent outcomes of the A-STEP 2030 (Attracting DiverSe Talent to the Engineering Professions of 2030) Erasmus+ project for preparing the engineers of tomorrow to meet SDG’s for a better world
Language-biased image classification: evaluation based on semantic representations
No full textInternational audienceHumans show language-biased image recognition for a word-embedded image, known as picture-word interference. Such interference depends on hierarchical semantic categories and reflects that human language processing highly interacts with visual processing. Similar to humans, recent artificial models jointly trained on texts and images, e.g., OpenAI CLIP, show language-biased image classification. Exploring whether the bias leads to interference similar to those observed in humans can contribute to understanding how much the model acquires hierarchical semantic representations from joint learning of language and vision. The present study introduces methodological tools from the cognitive science literature to assess the biases of artificial models. Specifically, we introduce a benchmark task to test whether words superimposed on images can distort the image classification across different category levels and, if it can, whether the perturbation is due to the shared semantic representation between language and vision. Our dataset is a set of word-embedded images and consists of a mixture of natural image datasets and hierarchical word labels with superordinate/basic category levels. Using this benchmark test, we evaluate the CLIP model. We show that presenting words distorts the image classification by the model across different category levels, but the effect does not depend on the semantic relationship between images and embedded words. This suggests that the semantic word representation in the CLIP visual processing is not shared with the image representation, although the word representation strongly dominates for word-embedded images
Conversational agents for fostering curiosity-driven learning in children
No full textInternational audienceCuriosity is an important factor that enables independent learning in children. Research suggests that it is also a competence that can be fostered by training specific metacognitive skills and information-searching behaviors. In this light, we develop a conversational agent that helps children generate curiosity-driven questions, and encourages their use to lead autonomous explorations and gain new knowledge. The study was conducted with 51 primary school students aged between 9 and 10 years old who interacted with either a neutral agent or an “incentive” agent that helped curiosity-driven questioning by offering specific semantic cues. Results showed a significant increase in the number and the quality of the questions generated with the “incentive” agent. This interaction also resulted in longer explorations and stronger learning progress.Together, our results suggest that the more our agent is able to train children’s curiosity-related metacognitive skills, the better they can maintain their information-searching behaviors and the more new knowledge they are likely to acquire
L'inclusion scolaire des élèves avec TSA : Prospection et analyse documentaire pour co-concevoir et valider des outils numériques favorables à l'inclusion scolaire en milieu ordinaire des élèves avec TSA
No full textInternational audienc
Physically-based sound synthesis software for Computer-Aided-Design of piano soundboards
No full textInternational audienceThe design of pianos is mainly based on empirical knowledge due to the lack of a simple tool that could predict sound changes induced by modifications of the geometry and/or the mechanical properties of the soundboard. We introduce the concept of Sound Computer-Aided Design through the framework of a program that is intended to simulate the acoustic results of virtual pianos. The calculation of the sound is split into four modules that compute respectively the modal basis of the stiffened soundboard, the string dynamics excited by the hammer, the soundboard dynamics excited by the string vibration, and the sound radiation. The exact resemblance between synthesis and natural sounds is not the primary purpose of the software. However, sound synthesis of real and modified pianos are used as reference tests to assess our main objective, namely to reflect faithfully structural modifications in the produced sound, and thus to make this tool helpful for both instrument makers and researchers of the musical acoustics community
Security Analysis: From model to system analysis
No full textInternational audienceThere is a wide range of security solutions on cyber-physical systems, most aimed at preventing an adversary from gaining access to the system. However, to make a cyber-physical system more resilient and discover possible attack scenarios, it is necessary to analyze systems by taking into account their interactions with their environment. Standard formal analysis approaches are based on a model of the system. From a quantitative and qualitative point of view, the results of these analyzes depends on the model abstraction relative to the system. Usually, property verification is performed with formulas expressed in specific logics such as LTL or CTL. One of the problems is the semantic gap between textual requirements and these formalisms. In a security context, attacker interests are also necessary to take into account in the properties expression, in addition to system requirements. In this article we propose an approach allowing to analyze a real cyber-physical system while taking into account the interests of an attacker and while reducing the semantic gap between the textual requirements and logic formulas. The proposed methodology relies on the property specification patterns and the specification of an interface related to the state of the deployed embedded software. The motivating example used in this article comes from an industrial partner included in a collaborative project
Seasonal Distribution of the Fin Whale (Balaenoptera physalus) in Antarctic and Australian Waters Based on Passive Acoustics
No full textInternational audienceThe fin whale is listed as globally vulnerable, with ongoing threats to their population, yet little is known about the distribution and movements of the Southern Hemisphere sub-species, Balaenoptera physalus quoyi . This study assesses fin whale distribution in the Southern Hemisphere analysing acoustic recordings from 15 locations in Antarctic and Australian waters from 2002 to 2019. A seasonal acoustic presence of fin whales in Antarctic waters from late austral summer to autumn (February to June) with long-term, consistent annual usage areas was identified at the Southern Kerguelen Plateau and Dumont d’Urville sites. In comparison, limited vocal presence of fin whales was observed at the Casey site. In Australian waters, fin whales were seasonally present from austral autumn to mid-spring (May to October) on east and west coasts, with a decadal pattern of acoustic presence observed at Cape Leeuwin, WA. Two migratory pathways are identified, from the Indian sector of Antarctica to the west coast of Australia and from the Pacific sector of Antarctica to the east coast of Australia. The identified seasonal distributions and migratory pathways provide valuable information to aid in monitoring the recovery of this vulnerable sub-species. We suggest the identified distribution and dispersal from the Southern Kerguelen Plateau and Dumont d’Urville sites to the west and east coasts of Australia respectively, as well as the spatial separation between Antarctic sites, provide preliminary evidence of separate sub-populations of the Southern Hemisphere sub-species of fin whale
Sonification d’Echogrammes : un projet Sciences & Arts pour la Pédagogie, la Médiation et la Recherche
No full textInternational audiencePour étudier la biodiversité marine (poissons, plancton, déplacement des masses d'eau, etc.), les scientifiques utilisent des outils de traitement du signal appliqués aux mesures des sondeurs. Ils représentent aussi visuellement, en temps réel ou différé, les signaux ultrasonores reçus. Ces représentations s'appellent échogrammes. Le projet ici présenté propose, en complément, de transformer les échogrammes en signaux sonores audibles pour les analyser par l'écoute et profiter ainsi des excellentes acuités temporelle et fréquentielle du système auditif. La sonification est dans un premier temps éprouvée dans des contextes artistiques, en mettant l’accent sur la pédagogie et la médiation auprès du grand public : création de pièces sonores par des étudiants en hydrographie/océanographie et en sciences et techniques du son, mise en place d'installations d’art numérique publiques, composition d’oeuvres symphoniques, etc. Ces diverses expériences participeront à l’élaboration, dans un second temps, d’un outil d’analyse sonore de la composition et de l'évolution des écosystèmes, dont la pertinence scientifique et la complémentarité vis-à-vis des méthodes visuelles et signalistes traditionnelles devront être validées
On some path-dependent SDEs involving distributional drifts
No full textInternational audienceIn this paper, we study (strong and weak) existence and uniqueness of a class of non-Markovian SDEs whose drift contains the derivative in the sense of distributionsof a continuous function