Archive Ouverte INSA Lyon
Not a member yet
75807 research outputs found
Sort by
Construction few-shot et fusion de graphes de connaissances temporels dynamiques atomiques à partir de textes
No full textNational audienceAvec la croissance rapide des données, extraire des connaissances de textes non structurés est devenu crucial pour l'analyse en temps réel, l'inférence temporelle et les mémoires dynamiques des agents. Pourtant, les graphes de connaissances traditionnels négligent souvent l'évolution des données, limitant leur adaptabilité. De plus, les approches récentes utilisant des modèles de langage en zero-shot ou few-shot, sans fine-tuning ni ontologies spécialisées, restent sujettes à l'instabilité et à une couverture incomplète des faits. Pour surmonter ces limites, nous proposons ATOM , une méthode few-shot et scalable pour construire et mettre à jour en continu des graphes de connaissances temporels à partir de textes non structurés. Notre approche segmente les documents en « faits atomiques » afin d'améliorer l'exhaustivité et la stabilité, génère des graphes temporels atomiques via une modélisation temporelle duale distinguant observation et validité, puis fusionne ces graphes en parallèle. Les résultats empiriques montrent des gains d'environ 18% en exhaustivité, 17% en stabilité, et plus de 90% de réduction de latence par rapport aux méthodes de référence
Automatic seal quality inspection using deep learning in mono material flexible packaging
No full textInternational audienceThe automated detection of seal flaws with thermal imaging represents one of the major problems in the manufacturing process. This paper suggests a thermal imaging based deep learning method of detecting sealing faults of mono-material flexible packaging. We compare multiple pre-trained Convolutional Neural Networks to detect more accurately and run faster. Three experimental tests are carried out to enhance the classification process. Another way of reducing model size and computational load with pruning and quantization is required to run it on edge devices. With high-performance NVIDIA GPUs, the adjusted models have an accuracy of 98.7\%, precision of 80.69\%, and recall of 88.89\%. This method performs effectively in real-time seal defect identification and may be applied in packaging quality checks in the industrial line. The model requires tuning to operate more effectively and adapt to a variety of production circumstances and packaging
Repairing and caretaking or managing and curating : how to build a "milieu" in the post-mining area of Saint-Etienne (France)
No full textInternational audienc
VLSF Decoding with Reliability Guarantees over Correlated Noncoherent Fading Channels
No full textThis report studies reliability-guaranteed decoding for variable-length stop-feedback (VLSF) codes over correlated noncoherent fading channels. The decoding rule is based on the evolution of the information density associated with a channel input-output realization. Due to channel memory, exact evaluation of this information density is intractable. To enable a constructive decoding, computable finite-blocklength lower and upper bounds on the information density that hold uniformly over time along each input-output sequence are derived. The lower bound enables a stopping-time analysis for VLSF decoding and has an operational meaning, while the upper bound quantifies the relaxation gap. Moreover, the relaxation gap between the bounds is explicitly characterized. For Gaussian signaling, the stopping-time distribution and the impact of fading correlation on decoding performance are numerically studied
Blocking Sphingosine 1‐phosphate Metabolism With Fingolimod Prevents the Progression of Vascular Smooth Muscle Cells Calcification in Chronic Kidney Disease
No full textInternational audienceABSTRACT Patients with chronic kidney disease, and particularly those under hemodialysis, are prone to develop cardiovascular complications, mostly due to the exacerbation of vascular calcification. Vascular calcification relies on the transdifferentiation of vascular smooth muscle cells into calcifying cells. Sphingosine 1‐phosphate is a pleiotropic sphingolipid and an important regulator of osteogenesis and the cardiovascular system. Therefore, we explored the role of sphingosine 1‐phosphate metabolism in chronic kidney disease‐derived vascular calcification. Vascular calcification progression in chronic kidney disease and sphingosine 1‐phosphate signaling were examined in calcified vascular smooth muscle cells, in aortic explants, in rats with adenine‐induced chronic kidney disease, as well as in serum from hemodialysis patients. Sphingosine kinase 2 activity and sphingosine 1‐phosphate secretion, under the control of phospholipase D1, were exacerbated in calcified vascular smooth muscle cells. Furthermore, phospholipase D1 knockout mice display significantly less circulating sphingosine 1‐phosphate, supporting intertwined signalization cascades. Overall, sphingosine kinase expression and activity were upregulated in calcified aortic explants and in calcified aortas from rats. Sphingosine 1‐phosphate was increased in the serum of rats with mild vascular calcification. The Food and Drug Administration‐approved immunosuppressant drug fingolimod, a general modulator of S1P metabolism, strongly inhibited calcification in vascular smooth muscle cells and aortic explants. Additionally, fingolimod significantly reduced inflammation, attenuated metabolic syndrome and moderately inhibited aortic calcification in rats. Finally, we demonstrated for the first time that serum sphingosine 1‐phosphate was significantly increased in hemodialysis patients with mild abdominal aortic calcification. Our findings open an unexplored therapeutic option, which is targeting sphingosine 1‐phosphate metabolism, eventually with fingolimod, for the prevention and treatment of vascular calcification in chronic kidney disease patients
Influence of electrical properties on thermal boundary conductance at metal/semiconductor interface
No full textInternational audienceRecent experimental investigations have demonstrated that doping a semiconductor is a route to increase the thermal boundary conductance at metal/semiconductor interfaces. In this work, the influence of the electrical properties on heat transfer across metal/doped semiconductor junctions is investigated. Specifically, thermal boundary conductance at the interfaces between p- and n-doped silicon and titanium is measured by employing frequency-domain photothermal radiometry under varying external conditions. The influence of the doping level of the semiconductor, the barrier height, and the space charge area is analyzed. In particular, a 40% increase in the interface thermal conductance with the application of a current at n-doped silicon/titanium interfaces is reported. The enhancement of the thermal boundary conductance is explained by the shrinking of the surface charga area induced by the electric current. This study opens the way to modulating interfacial heat transfer at metal/semiconductor interfaces through fine tuning of electrical effects
Mechanosensing and Sphingolipid-Docking Mediate Lipopeptide-Induced Immunity in <i>Arabidopsis</i>
No full textBacteria-derived lipopeptides are immunogenic triggers of host defenses in metazoans and plants. Root-associated rhizobacteria produce cyclic lipopeptides that activate systemically induced resistance (IR) against microbial infection in various plants. How these molecules are perceived by plant cells remains elusive. Here, we reveal that immunity activation in Arabidopsis thaliana by the lipopeptide elicitor surfactin is mediated by docking into specific sphingolipid-enriched domains and relies on host membrane deformation and subsequent activation of mechanosensitive ion channels. This mechanism leads to host defense potentiation and resistance to the necrotroph B. cinerea but is distinct from host pattern recognition receptor-mediated immune activation and reminiscent of damage-induced plant immunity. Main TextLipopeptides (LPs) represent a prominent and structurally heterogeneous class of molecules among the broad spectrum of small specialized metabolites synthesized by bacteria. Besides serving key functions for the ecological fitness of the producer (motility, biofilm formation, colonization, nutrient acquisition, or antagonism towards competing neighbors), some LPs also act as triggers of immune responses that restrict pathogen infection of metazoans and plants 1,2 . The vast majority of LPs formed by plant-associated bacteria are comprised of a partly or fully cyclized oligopeptide linked to a single fatty acid chain. Some of these cyclic lipopeptides (CLP) formed by beneficial species belonging to the Pseudomonas and Bacillus genera are potent elicitors of immune responses in the host plant leading to a systemically induced resistance (IR) against infection by microbial pathogens 2,3 .</div
LAD2025, A constraint-based solver for the subgraph isomorphism problem
No full textInternational audienceThe Subgraph Isomorphism Problem (SIP) is an N P-complete problem that aims at finding a copy of a pattern graph in a target graph. It may be modelled as a constraint satisfaction problem in a very straightforward way, and exact approaches for solving SIPs usually propagate constraints to reduce the search space. In particular, PathLAD is a solver introduced in 2016 that combines Locally All Different (LAD) constraints with path-based supplemental constraints. In this paper, we introduce LAD2025, which combines a complete refactoring of PathLAD with new features: new supplemental constraints, a weight-based variable ordering heuristic, random restarts with nogood recording, a new value ordering heuristic and a rule for selecting the level of filtering
Token positional games
No full textThe classical Maker-Breaker positional game is played on a board which is a hypergraph H, with two players, Maker and Breaker, alternately claiming vertices of H until all the vertices are claimed. When the game ends, Maker wins if she has claimed all the vertices of some edge of H; otherwise, Breaker wins. Playing this game in real life can be done by placing tokens on the vertices of the board. In this paper, we study the unfortunate case in which one or both players do not have enough tokens to cover all the vertices and, as such, will have to move their tokens around at some point instead of placing new ones. There may be a bias, in that Maker and Breaker do not necessarily have the same amount of tokens. The present paper initiates the study of this generalization of positional games, called token positional games.A particularly interesting case is when Maker has a winning strategy in the classical game: what is the lowest number of tokens with which she still wins against Breaker's unlimited stock? We notably show that, for k-uniform hypergraphs on an arbitrarily large number n of vertices, this number equals k if k ∈ {2, 3} but can vary from k to Ω(n) if k ≥ 4. From an algorithmic point of view, PSPACE-hardness in general is inherited from classical positional games, but we get a polynomial-time algorithm to solve the case where Breaker only has one token. We also establish EXPTIME-completeness for a "token sliding" variation of the game.</div
The Generation Phases of Flow Matching: a Denoising Perspective
No full textFlow matching has achieved remarkable success, yet the factors influencing the quality of its generation process remain poorly understood. In this work, we adopt a denoising perspective and design a framework to empirically probe the generation process. Laying down the formal connections between flow matching models and denoisers, we provide a common ground to compare their performances on generation and denoising. This enables the design of principled and controlled perturbations to influence sample generation: noise and drift. This leads to new insights on the distinct dynamical phases of the generative process, enabling us to precisely characterize at which stage of the generative process denoisers succeed or fail and why this matters