Scientific Publications of the University of Toulouse II Le Mirail
Not a member yet
92205 research outputs found
Sort by
LNA-Mixer Front-Ends Using Sampling Techniques for 77 GHz Automotive Radar Receivers
No full textInternational audienceThis paper investigates the use of sampling and subsampling techniques for frequency conversion in 77 GHz automotive radar receivers. Three LNA-mixer front-ends implementing these techniques are presented and compared. The first front-end implements direct sampling at 77 GHz using an innovative local oscillator (LO) pulse-shaper. Both other front-ends implement subsampling with a 25.7 GHz LO, significantly relaxing LO generation constraints while achieving better performance than conventional subharmonic mixers.All three front-ends were fabricated using a 28-nm FD-SOI CMOS process. The first design demonstrates a state-of-the-art noise / linearity trade-off while reducing power consumption by a factor of three compared to the best-performing CMOS and BiC-MOS solutions. It achieves an input-referred 1 dB compression point (ICP) of -7.8 dBm, a 8.9 dB noise figure (NF), and a 11.4 dB front-end gain, consuming only 20 mW at 1 V. The subsampling front-ends achieve an NF below 12 dB and ICP above -12 dBm. A complete description of these front-ends is provided, including the LO signal generation circuits and a theoretical analysis of their double-balanced mixer operation, considering the effects of LO transition time on conversion gain.</p
Détection de fraudes scientifiques: Détection de manipulations de citations par fouille de réseaux de citations
No full textLicenceLes scientifiques produisent et diffusent le savoir scientifique via différents médiums, les plus connus étant les articles scientifiques. Malheureusement, les chercheurs sont soumis à une pression des métriques pour avancer dans leur carrière : il leur faut publier plus, et toujours plus vite. Cette pression peut en pousser certains à pratiquer des méconduites scientifiques, comme la manipulation de citations, la falsification de données, le plagiat, et la vente d’articles...Ici, nous nous intéresserons particulièrement aux manipulations de citations : comment les caractériser ? Comment les détecter ?Pour cela, nous avons exploité 64To de méta-données scientifiques, fournies par Crossref, en utilisant une base de données Oracle (SQL) et des métriques que nous avons imaginées.Lien de la vidéo : https://www.youtube.com/watch?v=I66WXoSTPr
Analyzing Shapley Additive Explanations to Understand Anomaly Detection Algorithm Behaviors and Their Complementarity
No full textInternational audienceUnsupervised anomaly detection is a challenging problem due to the diversity of data distributions and the lack of labels. Ensemble methods are often adopted to mitigate these challenges by combining multiple detectors, which can reduce individual biases and increase robustness. Yet building an ensemble that is genuinely complementary remains challenging, since many detectors rely on similar decision cues and end up producing redundant anomaly scores. As a result, the potential of ensemble learning is often limited by the difficulty of identifying models that truly capture different types of irregularities. To address this, we propose a methodology for characterizing anomaly detectors through their decision mechanisms. Using SHapley Additive exPlanations, we quantify how each model attributes importance to input features, and we use these attribution profiles to measure similarity between detectors. We show that detectors with similar explanations tend to produce correlated anomaly scores and identify largely overlapping anomalies. Conversely, explanation divergence reliably indicates complementary detection behavior. Our results demonstrate that explanation-driven metrics offer a different criterion than raw outputs for selecting models in an ensemble. However, we also demonstrate that diversity alone is insufficient; high individual model performance remains a prerequisite for effective ensembles. By explicitly targeting explanation diversity while maintaining model quality, we are able to construct ensembles that are more diverse, more complementary, and ultimately more effective for unsupervised anomaly detection
Hierarchical Knowledge-Model-Induced Training for Object Detection: Enhancing Annotations with Class Variants
No full textHierarchy is a fundamental umbrella concept in visual systems used in different ways in Computer Vision. It is based on relations of various types between the elements that make up a scene, including generalization/specialization and composition/decomposition. In this work, we assume that combining these different relations can improve object detection perfomance. We propose a lightweight integration of hierarchy during both training and inference within an object detection Deep Learning architecture, specifically in the YOLO (You Only Look Once) [14] toolset. We introduce YOLO-KM, an extension of YOLO that associates sets of classes derived from the knowledge models referred to as variants with objects rather than simple classes. This contribution also addresses the adaptation of annotations and data processing across all stages of training and inference. We conduct a comparative study between standard YOLO and YOLO-KM for object and part detection, based on the widely used Pascal VOC, Pascal Part, and Semantic Pascal Part datasets. The results demonstrate that our approach improves YOLO's detection performance without compromising training or inference efficiency. Our work is available in a GitHub repository : https://github.com/ChristophePantel/ultralytics/tree/hierarchical.</div
Disclosing donor conception: A mixed methods study exploring the experience and attitudes of French sperm donor-conceived adults born within heterosexual couples
No full textInternational audienceResearch-question:How can we understand the experiences and attitudes of donor-conceived people regarding the disclosure of their conception, given the current emphasis on openness and the ongoing challenges faced by donor-conceived families?Design:This article draws on a mixed-method exploratory study (March 2019-September 2020) involving an online survey and in-depth interviews with French sperm donor-conceived adults born to heterosexual couples. The survey included 107 respondents aged 20-54, mostly women (86.9%) with high academic standards, informed of their conception at an average age of 18.6 (± 11.9). The interview involved 20 participants aged 21-53, most of whom were women (85%) and employed as managers (35%), informed between ages 5 and 49 (average: 19.7).Results:Age was not the primary factor for understanding how our donor-conceived participants experienced and perceived disclosure. Instead, the circumstances of disclosure emerged as an independent factor, with four paths identified: 1) disclosure as a mutually agreed parental strategy; 2) disclosure perceived as necessary due to events in the donor offspring’s life; 3) disclosure as a breach in a prior agreement between partners following a family event; 4) disclosure initiated by donor-conceived people themselves. Additionally, three factors explained the gendered gap that positioned mothers as the main actors in disclosure experiences: the organisation of MAR protocols, the social perception of male infertility, and the gendered division of labour within families.Conclusions:The study calls for a contextualised approach to information-sharing practices. Disclosure should be understood within its temporal and relational context to grasp how donor-conceived people experience learning the facts of their conception
PSPACE-completeness of bimodal transitive weak-density logic
No full textInternational audienceWindows have been introduced in \cite{BalGasq25} as a tool for designing polynomial algorithms to check satisfiability of a bimodal logic of weak-density. In this paper, after revisiting the ``folklore'' case of bimodal K4 already treated in \cite{Halpern} but which is worth a fresh review, we show that {\window}s allow to polynomially solve the satisfiability problem when adding transitivity to weak-density, by mixing algorithms for bimodal K4 together with windows-approach. The conclusion is that both satisfiability and validity are PSPACE-complete for these logics
: Monteray tradition and real appraoch: is the separation insurmontable?
No full textInternational audienceThis article aims to clarify the differences and similarities between the real and monetary approaches to economic theory, which Schumpeter and Cartelier identify as the two main branches. First, we demonstrate that the unsuccessful attempts to synthesise the work of Sraffa and Keynes, or Marx and Keynes, are undoubtedly linked to the irreconcilable nature of these authors' initial postulates and hypotheses. Sraffa and Marx, on the one hand, and Keynes, on the other, belong to opposing traditions. However, an analysis of certain works belonging to the real approach reveals affinities with the monetary approach. The asymmetry hypothesis between entrepreneurs and employees, analysis of imbalances and the status of money as a necessary intermediary in exchanges are not exclusive to the monetary tradition.L’objet de cet article est d’éclaircir ce qui sépare ou au contraire ce qui peut rapprocher les approches réelle et monétaire, qui, d’après Schumpeter et Cartelier, sont les deux grandes branches de la théorie économique. Dans un premier temps, on montre que les échecs relatifs des tentatives de synthèse entre Sraffa et Keynes ou encore Marx et Keynes sont sans doute liés au fait que les postulats et hypothèses de départ de ces auteurs ne sont pas conciliables. Sraffa et Marx d’un côté, Keynes de l’autre, appartiennent à deux traditions opposées. Cependant, l’étude de certains travaux relevant de l’approche réelle révèle des affinités avec l’approche monétaire. L’hypothèse d’asymétrie entre entrepreneurs et salariés, l’analyse des déséquilibres ou encore le statut de la monnaie comme intermédiaire nécessaire des échanges ne sont pas le monopole de la tradition monétaire
Transnationalisation des répertoires de motifs et d’actions fanoniens au Moyen-Orient et en Europe
No full textInternational audienceLa pensée de Fanon a eu une incidence notable sur nombre de mouvements révolutionnaires à travers le monde. Dans le présent chapitre, nous étudierons son influence sur deux mouvements de libération nationale, en l'occurrence basque et kurde (de Turquie). Nous envisagerons plus particulièrement la transnationalisation des répertoires de motifs et d'actions fanoniens. Nous nous appuierons sur deux enquêtes de sociologie qualitative réalisées dans chacune des zones géographiques auprès de militant.e.s et ex-militant.e.s clandestins. Nous montrerons en particulier comment, dans ces groupes, le référentiel marxiste a progressivement été remplacé par le paradigme fanonien des luttes de libération nationale et la promotion de ses répertoires d'action tels que l'usage de la violence politique. Afin de préciser ces transformations, nous envisagerons, dans un premier temps, les facteurs qui ont permis le dépassement des cadres théoriques du marxisme dans le « fanonisme ». Nous analyserons ensuite la façon dont les répertoires d'action théorisés par F. Fanon ont été appropriés sur les plans méso-et micro-sociaux, à partir notamment des vecteurs cognitifs et émotionnels de leur appropriation individuelle. Nous étudierons enfin le rôle de la référence à l'oeuvre de Fanon dans les transformations internes des organisations clandestines considérées sur le temps long
Dispositifs discursifs d’accompagnement et fabrique institutionnelle de l’hyper-vulnérabilité du patient incurable
No full textInternational audienceDans une perspective critique, cet article interroge la fabrique institutionnelle de l’hyper-vulnérabilité des patients atteints de cancer incurable. L’analyse communicationnelle proposée s’articule autour de trois axes : les rapports entre vérité, mensonge et vulnérabilité dans l’asymétrie symbolique du colloque médecin-patient ; la vulnérabilité du patient à l’épreuve des stratégies discursives du care et du cure ; le glissement de la vulnérabilité vers l’hyper-vulnérabilité lorsque la compétence du malade se trouve placée au coeur du pouvoir institutionnel. En conclusion, cette analyse suggère de concevoir l’hyper-vulnérabilité du malade non seulement comme produite par des dispositifs institutionnels d’accompagnement, mais aussi comme révélatrice des modalités contemporaines du lien social
Bronze Age copper supply in Mediterranean France: first results from lead isotope and chemical analyses of hoarded metalwork
No full textInternational audienceCopper supply networks in southern France are analysed on the basis of a study of five hoards of metal objects dating from the end of the Early Bronze Age (c. 17th-16th BCE) to the end of the Late Bronze Age (c. 9th BCE). A total of 73 inductively coupled plasma-optical emission spectroscopy elemental analyses were performed and 48 objects belonging to the different groups that could be identified from the elemental compositions were targeted for lead isotope analysis (multicollector-inductively coupled plasma mass spectrometry). The results clearly show that the same source was used for the various objects in each hoard, but that copper from different origins was used depending on the period. This reveals evolving supply networks that can be linked to the cultural interactions observed during this period. Towards the end of the Early Bronze Age or the beginning of the Middle Bronze Age (c. 17th-16th BCE), axe-ingots were imported into southern France along the Rhone corridor. The origin of the copper from which they are made could potentially be the Vosges massif. On the other hand, one of the major contributions of this study is to have demonstrated the use of copper originating from the Southeastern Alps during a late phase of the Middle Bronze Age and up to the Late Bronze Age (c. 14th-11th BCE). A form of packaging for this Alpine copper was as pick-ingots, and it was probably in this form that it arrived in southern France. This network was interrupted around the 9th century BCE and the populations of southern France returned to the exploitation of local minerals (Cabrières-Péret district). This research highlights the link between trade networks and cultural dynamics, showing that the circulation of raw materials also helps strengthen relations between communities