INRIA a CCSD electronic archive server
Not a member yet
122212 research outputs found
Sort by
Results of the Ontology Alignment Evaluation Initiative 2025
No full textInternational audienceThe Ontology Alignment Evaluation Initiative (OAEI) aims at comparing ontology matching systems on precisely defined test cases. These test cases can be based on ontologies of different levels of complexity and use different evaluation modalities. The OAEI 2025 campaign offered 12 tracks and was attended by 20 participants. This paper is an overall presentation of that campaign
Seed2LP : inférence de graines dans les réseaux métaboliques pour des applications d'écologie inverse
No full textNational audienceBackground: A challenging problem in microbiology is to determine nutritional requirements of microorganisms and culture them, especially for the microbial dark matter detected solely with culture-independent methods. The latter foster an increasing amount of genomic sequences that can be explored with reverse ecology approaches to raise hypotheses on the corresponding populations. Building upon genome scale metabolic networks (GSMNs) obtained from genome annotations, metabolic models predict contextualised phenotypes using nutrient information. Results: We developed the tool Seed2LP, addressing the inverse problem of predicting source nutrients, or seeds, from a GSMN and a metabolic objective. The originality of Seed2LP is its hybrid model, combining a scalable and discrete Boolean approximation of metabolic activity, with the numerically accurate flux balance analysis (FBA). Seed inference is highly customisable, with multiple search and solving modes, exploring the search space of external and internal metabolites combinations. Application to a benchmark of 107 curated GSMNs highlights the usefulness of a logic modelling method over a graph-based approach to predict seeds, and the relevance of hybrid solving to satisfy FBA constraints. An extension to communities of microorganisms is currently in development. Considering a community of networks increases the possibility of seed compounds, but also add a new component to the problem, defined by the transfers of nutrients between networks, increasing further the combinatorics of the problem. Conclusions : Focusing on the dependency between metabolism and environment, Seed2LP is a computational support contributing to address the multifactorial challenge of culturing possibly uncultured microorganisms. Seed2LP is available on https://github.com/bioasp/seed2lpContexte : L'un des défis majeurs en microbiologie consiste à déterminer les besoins nutritionnels des micro-organismes et à les cultiver, en particulier pour la matière noire microbienne détectée uniquement à l'aide de méthodes indépendantes de la culture. Ces dernières favorisent une augmentation du nombre de séquences génomiques pouvant être explorées à l'aide d'approches d'écologie inverse afin d'émettre des hypothèses sur les populations correspondantes. S'appuyant sur les réseaux métaboliques à l'échelle du génome (GSMN) obtenus à partir d'annotations génomiques, les modèles métaboliques prédisent des phénotypes contextualisés à l'aide d'informations sur les nutriments. Résultats : Nous avons développé l'outil Seed2LP, qui traite le problème inverse consistant à prédire les nutriments sources, ou graines, à partir d'un GSMN et d'un objectif métabolique. L'originalité de Seed2LP réside dans son modèle hybride, qui combine une approximation booléenne évolutive et discrète de l'activité métabolique avec l'analyse d'équilibre des flux (FBA) numériquement précise. L'inférence des graines est hautement personnalisable, avec plusieurs modes de recherche et de résolution, explorant l'espace de recherche des combinaisons de métabolites externes et internes. Son application à un benchmark de 107 GSMN sélectionnés met en évidence l'utilité d'une méthode de modélisation logique par rapport à une approche basée sur des graphes pour prédire les graines, ainsi que la pertinence de la résolution hybride pour satisfaire les contraintes FBA. Une extension aux communautés de micro-organismes est actuellement en cours de développement. La prise en compte d'une communauté de réseaux augmente le nombre de composés graines possibles, mais ajoute également une nouvelle composante au problème, définie par les transferts de nutriments entre les réseaux, ce qui augmente encore la combinatoire du problème. Conclusions : En se concentrant sur la dépendance entre le métabolisme et l'environnement, Seed2LP est un support informatique qui contribue à relever le défi multifactoriel de la culture de micro-organismes potentiellement non cultivables. Seed2LP est disponible sur https://github.com/bioasp/seed2l
When Mitigating Bias is Unfair: Multiplicity and Arbitrariness in Algorithmic Group Fairness
No full textInternational audienc
Mapping the Landscape of Dialogue Research: A Meta-Analysis of ACL Anthology 2024
No full textInternational audienceConversation is a foundational yet methodologically challenging object of study in computational linguistics. While dialogue has long been central to human communication, its multimodal, interactive, and context-dependent nature complicates both data collection and evaluation. In this meta-analysis, we examine how conversation is currently conceptualised and studied within the NLP/CL community by analysing dialogue-related papers published in 2024 across major venues indexed in the ACL Anthology. Using metadata retrieved via the ACL Anthology API and manual annotation, we identify 445 papers addressing conversational phenomena. We analyse the conversational tasks studied, the corpora employed, and the evaluation methods used. Our findings show a strong dominance of task-oriented dialogue systems, with widespread reliance on a limited set of benchmark datasets and on automatic reference-based evaluation metrics. We highlight recurring issues, including insufficient corpus documentation, the use of scripted or synthetic dialogue as a proxy for natural conversation, and evaluation practices that inadequately capture conversational quality. We argue for the need to make task-specific expectations of conversation explicit and to develop evaluation frameworks that better reflect the richness and diversity of conversational interaction
Circadian Clock Model with Sequestration Repression Motif: Existence of Periodic Orbits and Entrainment Properties
No full textInternational audienc
Guided Clustering Variational Autoencoder (GCVA): A Predictive Clustering Approach with a VAE
No full textInternational audienc
Neural signature of indirect reply processing while listening to foreign accented dialogues
No full textInternational audienceThe interpretation of the sentence ‘Giving a good presentation is complicated’ will differ whether it is a reply to (a) or (b):(a) What is it like giving a presentation?(b) Did you like my presentation?The neurocognitive mechanisms underlying the pragmatic inference of indirect replies (e.g. in reply to (b)), have mainly been studied when dialogues occur in first language contexts. However, we hypothesize that foreign-accented speech may affect such inferences given the cognitive cost it generates (due to linguistic disfluency), as well as native listeners’ limited expectation of the foreign speaker’s linguistic abilities. To test this hypothesis, our ongoing experiment aims to characterize the neural signature of indirect reply processing in foreign-accented dialogues. Accordingly, we measure the event-related potentials (ERPs) of 40 French-speaking students listening to native and foreign-accented dialogues. Each dialogue is preceded by a written context establishing the communicative situation. A total of 144 dialogues are presented, ending in either a direct or indirect reply (e.g. in reply to (a) or (b), respectively). For one-third of the dialogues, participants answer yes-no comprehension questions.Based on the few previous ERP studies of indirect replies and foreign-accented pragmatics, we expect to find ERPs time-locked to the final word of the reply that reflect additional cognitive effort when processing foreign-accented indirect replies. Compared to native-accented direct replies, we anticipate a shallower processing. Particularly, we should observe a smaller and delayed N400, indicating a hindered semantic integration of the reply, and a larger P600, reflecting the additional pragmatic processing. Moreover, we will examine explanatory hypotheses of how individual cognitive capacities, such as working memory and non-verbal reasoning, may modulate pragmatic processing in multilingual communication, as previously observed in native-accented contexts. Our results will be discussed in line with the neurocognitive models of language comprehension and pragmatics
A Robust Discrete-Time Internal Model-based Controller for Continuous-Time LTI Systems with Uncertain Periodic Disturbances
No full textInternational audienceThis paper introduces a robust discrete-time controller that combines internal model-based control with sliding mode techniques to reject periodic and matched disturbances in linear time-invariant systems. By explicitly accounting for discretization effects inherent in digital implementations, the controller bridges continuous-time dynamics and sampled-data operation. A systematic design approach is developed to handle uncertainties in the exo-system, ensuring bounded tracking error and improved disturbance rejection. Simulation studies on an active suspension system illustrate that the proposed method outperforms traditional internal model-based and sliding mode controllers, yielding reduced overshoot and tighter error bounds
A Modular Orthogonal Integration of Operational and Prescriptive Timing Requirements using TASTD ⋆
No full textInternational audienceDesigning and verifying safety-critical systems requires giving equal importance to nonfunctional and functional requirements from the start, especially those related to time. Since no single method can address all system aspects, a combination of formal methods is needed for accurate modeling and analysis. This paper proposes an integration of the CCSL (Clock Constraint Specification language) and TASTD specification languages for the modular modeling of timing requirements. CCSL allows for the modular expression of timing constraints, whereas TASTD allows for the combination of state machines using CSP process algebra operators, to foster the construction of a system specification using the composition of small components specification. We propose rules to translate a CCSL specification into a TASTD specification, which can use the CSP operators of TASTD to compose the CCSL constraints with other specification elements, like operating modes of a system. The CCSL part of the specification can be verified using CCSL tools. The global specification can be simulated using TASTD tools. The approach is demonstrated on a temperature control system which has been previously modeled using UML MARTE and CCSL, but for which no formal combination was available
Positive Univariate Polynomials: SOS certificates, algorithms, bit complexity, and T-systems
No full textWe consider certificates of positivity for univariate polynomials with rational coefficients that are positive over (an interval of)~.Such certificates take the form of weighted sums of squares (SOS) of polynomials with rational coefficients.We build on the algorithm of Chevillard, Harrison, Jolde{\c{s}}, and Lauter~\cite{chml-usos-alg-11}, and we introduce a variant that we refer to as \usos. Given a polynomial of degree~ with maximum coefficient bitsize~, we show that \usos computes a rational weighted SOS representation in bit operations; the resulting certificate of posivitity involves rationals of bitsize . This improves the best-known complexity bounds by a factor of~ and completes previous analyses. We also extend these results to certificates of positivity over arbitrary rational intervals, via a simple transformation. In this case as well, our techniques yield a factor- improvement in the complexity bounds.Along the same line, for univariate polynomials with rational coefficients, we introduce a new class of certificates, which we call \emph{perturbed SOS certificates}.They consist of a sum of two rational squares that approximates the input polynomial closely enough so that nonnegativity of the approximation implies the nonnegativity of the original polynomial.This computation has the same bit complexity and yields certificates of the same bitsize as in the weighted SOS case.We further investigate structural properties of these SOS decompositions.Relying on the classical result that any nonnegative univariate real polynomial is the sum of two squares of real polynomials, we show that the summands form an interlacing pair. Consequently, their real roots correspond to the Karlin points of the original polynomial on~, establishing a new connection with the T-systems studied by Karlin~\cite{Karlin-repr-pos-63}. This connection enables us to compute such decompositions explicitly. Previously, only existential results were known for T-systems. We obtain analogous results for positivity over , and hence over arbitrary real intervals.Finally, we present our open-source Maple implementation of the \usos algorithm, together with experiments on various data sets demonstrating the efficiency of our approach