HAL Portal UTC Université de Technologie de Compiègne
Not a member yet
11652 research outputs found
Sort by
Credal discrete classifier
No full textInternational audienceThis paper presents a novel extension of the discrete Bayesian classifier (DBC) into a set-valued classification framework grounded in imprecise probability theory. The standard DBC framework, which relies on partitioning the input space into profiles and estimating class-conditional probabilities, may not be very robust to distribution changes or imperfections in observed data. In the hope to mitigate such issues, we introduce the Credal Discrete Classifier (CDC), an imprecise-probabilistic extension of the traditional Bayesian approach. By representing uncertainties in the estimated probabilities through belief functions, CDC offers interval-valued risks and set-valued decisions, thereby enhancing robustness. Experimental results on several benchmark datasets demonstrate that CDC effectively balances accuracy and determinacy by allowing for set-valued predictions in uncertain contexts, often outperforming or matching traditional precise classifiers
From lab to market: real-time food safety monitoring via spectroscopy, blockchain and Artificial Intelligence
No full textInternational audienceOilseed rape (Brassica napus), one of the most important sources of vegetable oil worldwide, is adversely impacted by heat wave-induced temperature stress especially during its yield determining reproductive stages. However, the underlying molecular and biochemical mechanisms are still poorly understood. In this study, we investigated the transcriptomic and metabolomic responses to heat stress in B. napus plants exposed to a gradual increase of temperature reaching 30 °C in the day and 24 °C at night for a period of 6 days. High Performance Liquid Chromatography (HPLC) and Liquid Chromatography-Mass Spectrometry (LC-MS) was used to quantify the content of carbohydrates and glucosinolates respectively. RNA-Seq analysis of flower buds showed a total of 1892, 3253, 4553, 4165 and 1713 genes were differentially expressed at days 0, 1, 2, 6 and 12 after treatment respectively. Results showed that heat stress reduced yield and altered oil composition. Heat stress also increased the content of carbohydrate (Glucose, Fructose and Sucrose) and aliphatic glucosinolates (Gluconapin, Progoitrin) in the leaves but decreased the content of the indolic glucosinolate (Glucobrassicin). Heat treatment resulted in down regulation of genes involved in respiratory metabolism namely Glycolysis, Pentose Phosphate Pathway, Citrate Cycle and Oxidative Phosphorylation especially after 48 hrs of heat stress. Other down regulated genes mapped to sugars transporters, nitrogen transport and storage, cell wall modification and methylation. In contrast, up regulated genes mapped to small heat shock proteins (sHSP20) and other heat shock factors that play important roles in thermotolerance. Furthermore, two genes were chosen from the pathways involved in the heat stress response to further examine their expression using real time RT-qPCR. Taken together, the results of this study demonstrated that the changes in carbohydrates and glucosinolates metabolism and transport under heat stress is one of the response mechanisms employed by the plant under heat stress conditions. Moreover, the transcriptomic data will be useful for further understanding the molecular mechanisms of B. napus plant tolerance to heat stress and provide a basis for breeding heat-tolerant varieties
Evaluation of image quality parameters in lung using MR elastography technique
No full textInternational audienc
Les LLMs sur les traces de Voltaire: débusquer les anecdotes dans la presse du XVIIIᵉ siècle
No full textInternational audienc
Enhancing the valorization of Hemp Inflorescences using Instant Controlled Pressure Drop (D.I.C) Pretreatment
No full textInternational audienceObjectivesHemp (Cannabis sativa L.) has high potential for the valorization of its inflorescences, which are rich in bioactive and volatile compounds such as terpenes, cannabinoids, and polyphenols. The preservation and recovery of these molecules strongly depend on the drying process, which must be adapted to the sensitivity and volatility of the target compounds. Pretreatment using Instant Controlled Pressure Drop (D.I.C) technology induces expansion of the plant matrix through swelling and controlled decompression, which can significantly enhance drying efficiency and the subsequent recovery of volatile compounds. This study aims to evaluate the impact of D.I.C on drying kinetics and on the preservation and extraction of volatile components.Methods The cannabis inflorescences were provided by La Ferme Bio de Pigerolles (France) and frozen prior to D.I.C pretreatment, performed on pilot-scale equipment from ABCAR D.I.C Process (France). The D.I.C process involves a rapid increase in pressure by steam injection, a short holding period, and an instant controlled decompression. The controlled parameters are: • Pressure • Treatment duration • Number of cycles The pretreated and untreated samples were dried under identical conditions. Moisture content was monitored over time to determine drying curves and the characteristic drying time (τd). The structural modifications induced by D.I.C were observed under optical microscopy, and volatile compounds were analyzed to evaluate its impact on reducing volatile losses and on the composition of the plant’s volatile extracts. In addition, a theoretical study on the solubility of cannabidiol (CBD) in supercritical fluids—pure CO₂ and CO₂ mixed with an ethanol fraction—was conducted to predict optimal extraction parameters. This study aims to guide the design of upcoming fractionated extraction experiments of Cannabis sativa compounds using supercritical CO₂ with and without a co-solvent.ResultsD.I.C pretreatment significantly influenced the drying kinetics, reducing the characteristic drying time and overall drying duration. Microscopic observations confirmed the structural changes, while preliminary analyses indicated that D.I.C facilited internal mass transfer and release of essential oils. The volatile fraction recovered directly during D.I.C treatment (1.65%) was comparable to that obtained by hydrodistillation of frozen material (1.7%), whereas the residual D.I.C waste yielded only 0.05%, confirming the efficiency of volatile recovery during the process. Furthermore, fractionation of the volatile extracts revealed distinct chemical profiles: the fraction collected at the pump outlet was dominated by monoterpenes, whereas the fraction condensed inside the chamber was richer in sesquiterpenes, demonstrating the potential of D.I.C for controlled volatile separation.Theoretical modeling based on Hansen solubility parameters predicted optimal supercritical CO₂ extraction conditions around 80 °C and 200 bar for most compounds. However, for CBDA, the best solubility without a co-solvent was obtained at 70 °C and 250 bar, suggesting slightly different interaction behavior compared to the other cannabinoids.ConclusionThese results highlight the strong influence of D.I.C pretreatment on drying behavior and volatile compound recovery, as well as the relevance of theoretical modeling to guide extraction optimization. Based on the predicted solubility parameters, the next step will involve conducting supercritical CO₂ extraction experiments to identify the best compromise between CBD/CBDA yield and extract quality
Comparative analysis of analytical methods for the quantification of sulfates in reconstituted soil matrices: Effect of soil texture and interfering species
No full textInternational audienceThis article studies the effect of various soil textures on sulfate determination and compares the performance of ion chromatography and the turbidimetric method for quantifying sulfates in reconstituted soil matrices. The comparison is conducted under both interference-free conditions and in the presence of interferences, such as chlorides and fluorides. These interferences caused significant overestimations of theoretical sulfate concentrations when using the standardized turbidimetric method. For instance, at a theoretical sulfate concentration of 1000 mg/kg, the result obtained through the turbidimetric method was 12085.5 mg/kg in the presence of chlorides, while the result obtained through ion chromatography was 915.5 mg/kg. An alternative method relying on hyperspectral imaging is also presented for direct sulfate analysis in soil without the need for extraction. Hyperspectral analyses proved to enable rapid analysis and accurate prediction of sulfate concentrations in soil samples, with an R2 score of 0.87 in the absence of interferences and an R2 score of 0.9 in the presence of chlorides
Lower Bounds for the Cumulative Scheduling Problem Based on Energetic Reasoning
No full textInternational audienceThis paper addresses the Resource-Constrained Project Scheduling Problem (RCPSP) by decomposing it into multiple Cumulative Scheduling Problems (CuSPs), each handling resource constraints independently. The CuSP involves scheduling non-preemptive tasks within fixed time windows while respecting resource limitations. We focus on determining lower bounds for the minimal makespan in CuSP and present both destructive and constructive approaches while exploring the impact of redundant resources. \textcolor{black}{To achieve this, we leverage energetic reasoning, an advanced filtering technique for the Cumulative Scheduling Problem. While the classical definition of energetic reasoning assesses the portions of tasks processed inside a given time interval, our refined approach instead evaluates the portions processed outside the interval. This alternative perspective allows us to exploit the constraint that, immediately before and after the interval, resource availability is limited by an integer . By formulating this as a tripartition problem and solving it via Dynamic Programming, we improve the evaluation of scheduling intervals, leading to a more effective constraint propagation.} Experimental results demonstrate the effectiveness of our approach in optimizing project scheduling
Gallica-Obvie: un outil pour la valorisation et l'exploration du patrimoine textuel
No full textInternational audienceAvec ses 10 millions de documents, Gallica, la bibliothèque numérique de la Bibliothèque nationale de France (BnF), figure parmi les collections en ligne les plus vastes au monde. Enrichie d'outils d'analyse et de fouille textuelle, elle offre aux chercheurs en sciences humaines et sociales des perspectives inestimables pour explorer et interroger des corpus patrimoniaux dans un environnement numérique.Nous présentons Gallica-Obvie, un nouveau prototype permettant de connecter Gallica 1 à la plateforme Obvie 2 , un moteur de recherche doté de fonctionnalités avancées pour la fouille de textes, les statistiques lexicales et la comparaison de corpus. Gallica-Obvie permet aux chercheurs d'importer directement des documents depuis Gallica grâce à leurs identifiants ARK (Archival Resource Key). Une fois importés, ces documents sont automatiquement intégrés dans Obvie, où ils peuvent être indexés et analysés.Le processus, illustré par ce diagramme simplifié, commence par la saisie d'une liste de liens ARK dans l'interface utilisateur. L'outil interroge alors l'API de Gallica pour récupérer les documents demandés. Ces derniers sont transférés vers Obvie, où une instance temporaire du corpus est créée, immédiatement accessible en ligne. Cette instance offre une gamme d'outils d'analyse, notamment des recherches par mots clés lemmatisés, un filtrage précis par métadonnées, des concordances, des comparaisons inter-corpus et des visualisations interactive
Possibilistic Logic and Inference for Linear Systems
No full textInternational audienceIn this work, we propose using possibility theory to represent a belief base and to reason over it, particularly in the presence of inconsistencies. We revisit semantics from the existing literature and introduce two additional properties that help to clarify how these semantics are interconnected. The framework is then restricted to hypotheses expressed as linear numerical inequalities, in order to benefit from the polynomialtime complexity of Linear Programming while maintaining possibilistic reasoning. Finally, we define certified inference syntaxes based on Farkas' Lemma, allowing a certificate to be provided for each inference
Contribution à une méthodologie d'éco-conception proactive : état de l'art sur le PLM, la Circularité et la Soutenabilité
No full textInternational audienceContribution à une méthodologie d'éco-conception proactive : état de l'art sur le PLM, la Circularité et la Soutenabilit