HAL-INSA Toulouse
Not a member yet
34325 research outputs found
Sort by
Limiter le cyberharcèlement : une approche basée sur les modèles de Localisation-Interdiction
No full textInternational audienceIntroduction.Cette recherche s'inscrit dans la problématique du cyberharcèlement, un phénomène qui se définit comme « un acte agressif, intentionnel perpétré par un individu ou un groupe d'individus au moyen de formes de communication électroniques, de façon répétée à l'encontre d'une victime qui ne peut facilement se défendre seule » [8]. Notre objectif est de proposer une approche originale permettant d'identifier les liens à supprimer dans un réseau social afin de limiter au maximum la propagation du cyberharcèlement. Pour ce faire, nous nous appuyons sur des modèles issus d'un domaine particulier de la localisation optimale, à savoir le champ de la « Localisation-Interdiction » [7]
In-context learning for the classification of manipulation techniques in phishing emails
No full textTraditional phishing detection often overlooks psychological manipulation. This study investigates using Large Language Model (LLM) In-Context Learning (ICL) for fine-grained classification of phishing emails based on a taxonomy of 40 manipulation techniques. Using few-shot examples with GPT-4o-mini on real-world French phishing emails (SignalSpam), we evaluated performance against a human-annotated test set (100 emails). The approach effectively identifies prevalent techniques (e.g., Baiting, Curiosity Appeal, Request For Minor Favor) with a promising accuracy of 0.76. This work demonstrates ICL's potential for nuanced phishing analysis and provides insights into attacker strategies
Théorie de l'homotopie des opérades courbées et des algèbres courbées
No full textCurved algebras are algebras endowed with a predifferential, which is an endomorphism of degree −1 whose square is not necessarily zero. This makes the usual definition of quasi-isomorphism meaningless and therefore the homotopical study of curved algebras cannot follow the same path as differential graded algebras.In this article, we propose to study curved algebras by means of curved operads. We develop the theory of bar and cobar constructions adapted to this new notion as well as Koszul duality theory. To be able to provide meaningful definitions, we work in the context of objects which are filtered and complete and become differential graded after applying the associated graded functor. This setting brings its own difficulties but it nevertheless permits us to define a combinatorial model category structure that we can transfer to the category of curved operads and to the category of algebras over a curved operad using free-forgetful adjunctions. We address the case of curved associative algebras. We recover the notion of curved A∞-algebras, and we show that the homotopy categories of curved associative algebras and of curved A∞-algebras are Quillen equivalent
Wideband Tunable Filter of Dual-path Microstrip Coupled-lines with Varactor Tuned Circuit
No full textInternational audienceThis study presents a compact and tunable microstrip of dual-path wideband filter that employs coupled-lines and varactors to address the needs of 4G/sub-6 GHz 5G communication systems. This work integrates tunable methods inspired by wideband parallel coupled line-based topologies to realize a reconfigurable solution achieving a 34% frequency tuning range (1.13-1.51 GHz) while maintaining a low insertion loss of approximately below 1 dB. Specifically, the proposed microstrip-based filter, which is designed, uses parallel coupled-line resonators with a quarter-wavelength length, enabling a broad tuning range between 1.27-1.54 GHz. Adjusting the coupling strengths of both adjacent and non-adjacent resonators, the filter can be shifted within this frequency band without compromising performance. Therefore, to achieve the desired level of control, two identical varactor diodes and biasing circuitry are meticulously selected for their stable and repeatable capacitance-voltage characteristics to adjust the filter's resonant frequency. The optimal positions for these tuning circuits are determined based on the resulting capacitance, which is crucial for achieving a wide tuning range. Simulation and measurement confirm that this reconfigurable microstrip filter, implemented on a 60.7 × 35.4 mm 2 footprint, benefits not only from a reduced footprint but also from the ability to target multiple frequency bands with minimal hardware modifications, delivering the intended performance for modern wireless front ends
Enhancing Biodiversity‐Function Relationships in Field Retting: Towards Key Microbial Indicators for Retting Control
No full textInternational audienceHemp field retting is a bioprocess that facilitates fibre extraction by degrading pectin and other matrix components surrounding fibre bundles. However, traditional methods rely on empirical practices, often resulting in inconsistent fibre quality. This study investigates the biodiversity–function relationship in the hemp retting ecosystem to identify microbial and enzymatic indicators for improved process control. Over six weeks of field retting, we monitored bacterial and fungal community dynamics using high‐throughput sequencing and assessed enzymatic activity profiles. Our results revealed a sequential enzymatic pattern: pectinases (e.g., polygalacturonase) dominated early stages, followed by hemicellulases (β‐xylosidase, β‐galactosidase), and later cellulases. These enzymatic shifts were reflected in the changes in microbial community composition, with pectinolytic bacteria (e.g., Proteobacteria ) dominating the initial phases and cellulolytic fungi (e.g., Ascomycota ) becoming more prevalent later. Our results identified specific microbial taxa correlated with optimal retting, suggesting their potential as bioindicators for monitoring retting. Specifically, key bacterial genera such as Pseudomonas and Sphingomonas , and fungal genera like Cladosporium , were associated with distinct enzymatic profiles. Our findings offer new insights into the microbial ecology of retting, providing both microbial and enzymatic indicators that could inform the development of monitoring strategies for process control, ultimately contributing to more consistent hemp fibre production
Characteristics analysis of silicone gel electrical treeing for power device packaging under high temperature coupled square wave pulse electric field
No full textInternational audienc
Inference post region selection
No full textPost-selection inference consists in providing statistical guarantees, based on a data set, that are robust to a prior model selection step on the same data set. In this paper, we address an instance of the post-selection-inference problem, where the model selection step consists in selecting a rectangular region in a spatial domain. The inference step then consists in constructing confidence intervals on the average signal of this region. This is motivated by applications such as genetics or brain imaging. Our confidence intervals are constructed in dimension one, and then extended to higher dimension. They are based on the process mapping all possible selected regions to their corresponding estimation errors on the average signal. We prove the functional convergence of this process to a limiting Gaussian process with explicit covariance. This enables us to provide confidence intervals with asymptotic guarantees. In numerical experiments with simulated data, we show that our coverage proportions are fairly close to the nominal level already for small to moderate data-set size. We also highlight the impact of various possible noise distributions and the robustness of our intervals. Finally, we illustrate the relevance of our method to a segmentation problem inspired by the analysis of DNA copy number data in cancerology
In-situ piezoelectric sensors for structural health monitoring with machine learning integration
No full textInternational audienc
Zeeman splitting observations in laser-produced magnetized blast waves
No full textInternational audienceWe report the observation of Zeeman splitting in multiple spectral lines emitted by a laser-produced, magnetized plasma (1–3 × 1018 cm−3, 1–15 eV) in the context of a laboratory astrophysics experiment under a controlled magnetic field up to 20 T. Nitrogen lines (NII) in the visible range (563–574 nm) were used to diagnose the magnetic field and plasma conditions. This was performed by coupling our data with the Stark–Zeeman line-shape code PPPB. The excellent agreement between experiment and simulations paves the way for a non-intrusive experimental platform to get time-resolved measurements of the local magnetic field in laboratory plasmas