Archive ouverte HAL-LAAS
Not a member yet
12189 research outputs found
Sort by
TOWARDS UNVEILING THE HIDDEN DYNAMICS OF BACTERIORHODOPSIN WITH MEMS-BASED ATOMIC FORCE MICROSCOPY
No full textInternational audienceUnderstanding membrane protein unfolding is essential for elucidating protein stability and function.Here, we employ MEMS-based atomic force microscopy (AFM) to investigate the unfolding dynamics of bacteriorhodopsin (BR) with microsecond resolution. Compared to conventional cantilevers, High frequency MEMS probes minimize meniscus effects and enable highsensitivity force spectroscopy under dry conditions. Unfolding force-extension curves reveal discrete structural transitions consistent with BR domain unfolding. These findings demonstrate MEMS-AFM as a valuable tool for high-speed biomolecular force measurements. This study also lays the groundwork for implementing optomechanical probes at > 100 MHz, advancing timeresolved protein mechanics into the nanosecond regime
Le projet « femmes géographes » sur Wikipédia : analyse d'une participation à un commun du savoir
No full textInternational audienc
Sequential Sample Average Majorization–Minimization
No full textInternational audienceMany statistical inference and machine learning methods rely on the ability to optimize an expectation functional, whose explicit form is intractable. The typical method for conducting such optimization is to approximate the expected value problem by a size-N sample average, often referred to as sample average approximation (SAA) or M-estimation. When the solution to the SAA problem cannot be obtained in closed form, the Majorization--Minimization (MM) algorithm framework constitutes a broad class of incremental optimization solutions, relying on the iterative construction of surrogates, known as majorizers, of the original problem. The ability to solve an SAA problem depends on the availability of all N observations, contemporaneously, which is difficult when N is large or data are observed as a stream. We propose a stochastic MM algorithm that solves the expected value problem via iterative SAA majorizer constructions using sequential subsets of data, which we call Sequential SampleAverage Majorization–Minimization SAM2. Compared to previous stochastic MM algorithm variants, our method permit an extended definition of majorizers, and does not rely on convexity assumptions, smoothness assumptions, or restrictions on functional classes for objectives and majorizers. We develop a theory of stochastic convergence for SAM2, made possible via the presentation of a novel double array uniform strong law of large numbers. Examples of SAM2 algorithms are given along with a numerical demonstration of SAM2 to quantile regression problems, in the regular and sparse parameter settings, including both convex and non-convex objective functions
Integrating the Technical Level into a Model-Based Safety and Security Analysis: Why it's Necessary and How it Can be Done
No full textInternational audienceWe motivate why it is necessary to integrate the technical level into a model-based safety and security analysis on the dynamical system level, and show how this can be done
One-size-fits-all Evaluation of LLMs for Safety Assurance Considered Harmful
No full textInternational audienceLarge Language Models (LLMs) have been proposed to support the development and maintenance of assurance cases (ACs), but their use comes with risk. This position paper calls on academic and industrial interest holders to establish methodological standards to evaluate the application of LLMs to ACs. Our position is that there is no "one size-fits-all" evaluation method, and that evaluation must be tailored to the specific objectives and risk profiles of different LLM use cases. As a first step, we outline a preliminary taxonomy of characteristics, risk, and evaluation methods for LLM use cases
Partitioning of AI Models for Execution on Mixed Criticality Systems. A Workflow Approach Proposal
No full textInternational audienceWe anticipate that in dependable edgeAI AI system development, only a subset of AI-inferred classes must be inferred dependably. The implication is a mixed-criticality AI execution environment where critical processing can be treated substantially differently than any non-critical processing from which is follows that the AI model must be partitioned, allowing separation of critical and non-critical AI-execution. There is currently no standard method of achieving and documenting such a state. This paper examines partitioning for mixedcriticality execution
A modular risk concept for complex systems
No full textInternational audienceThis paper motivates the views that for complex systems, risk should be controlled by enforcing constraints in a modular way at different system levels, that the constraints can be expressed as assurance contracts and that acceptable risk mitigation can be demonstrated in assurance case modules. This short paper explains how already existing methodologies can be combined to create a concept for modular risk assessment. The main novelty is the use of so-called contract-based design (CBD) contracts and refinements as risk constraints. This idea is presented here with the objective of receiving feedback from industry and academia
Section 06 Sciences de l’information : fondements de l’informatique, calculs, algorithmes, représentations, exploitations: Rapport de conjoncture 2024
No full textLa section 6 du Comité national de la recherche scientifique est, avec la section 7, une des deux sections traitant de la science informatique, et plus précisément de l’algorithmique et de la combinatoire, du calcul, du logiciel, de la sécurité, des réseaux et systèmes distribuées, des données et connaissances, de l’intelligence artificielle et de l’aide à la décision, ainsi que de la bio-informatique et de l’informatique quantique. Ce rapport présente le périmètre thématique de la section, discute de la place des femmes ainsi que des évolutions récentes des pratiques de recherche, présente la conjoncture des différents thèmes de recherche et enfin décrit les carrières au CNRS des chercheurs et chercheuses de la section
Effective Data Generation and Feature Selection in Learning for Planning
No full textInternational audiencePrevious studies have shown that leveraging data beyond optimal training plans improves the learning of search guidance for planning. Specifically, state ranking information can be extracted from states on optimal plan traces and their siblings. In this paper, we generalise this approach by extracting additional rankings from the A⋆ search tree for generating optimal training plans. As in the previous approach, we incur no additional search effort and negligible computational overhead for data extraction. However, extracting more data in this way may introduce many redundant features and states which slows down training. We formalise the problem of sound, redundant feature pruning and show that it is NP-complete to solve. Furthermore, we introduce several algorithms and approximations for redundant feature pruning. Experiments show that rankings learned by extracting more data from search trees for generating optimal training plans improve planner coverage. However, pairing with unsound pruning methods often results in diminishing performance, while our sound feature pruning methods provide consistent improvements across tested domains
Temporal dynamics of natural sounds representation in the human brain
No full textAcoustic and semantic representations involved in the temporal dynamics of the cerebral processing of natural sounds are often studied separately. As a consequence, we lack direct knowledge of how the human brain transforms complex acoustic waveforms into semantic representations of the acoustic environment. Here, we aimed to elucidate this process by predicting magnetoencephalographic (MEG) responses to natural sounds using acoustic, and semantic (text-based) models. Critically, we also consider two recently developed soundprocessing convolutional neural networks (CNNs) that differ only in their loss function: CatDNN, which learns sound-event categories, and SemDNN, which learns continuous semantic embeddings (Word2Vec). We observe that DNNs better predict the dynamic MEG response, except at a long latency (800-1000 ms) where higher-level acoustics seems to dominate (auditory dimensions). Focusing on DNNs, we observe a potential switch from initial protoacoustic/categorical semantic representations (CatDNN, 250 ms) to more refined continuous semantic representations (SemDNN, 500-800 ms). Overall, our findings suggest limitations in the text-based modeling of the cerebral representations of natural sounds, and give a temporally resolved description of the cerebral dynamics of the acoustic-to-semantic transformation