Université de technologie de Troyes open archive
Not a member yet
10722 research outputs found
Sort by
A systemic model to assess shortage and wastage levels in demand-driven blood supply chains
No full textInternational audienceThe blood supply chain is complex and subject to uncertain supply and demand. This paper proposes a system dynamics model that considers the correlation between the supply of different blood products. The model evaluates the long-term effect of disturbances on the shortage and wastage levels of blood. The real case of Bogota (Colombia) is used for the analysis. The results of the experiments for a demand-driven policy provide managerial insights for limiting the expected shortage and wastage levels, especially when the supply of products is correlated. The proposed model is a tool for decision-makers to estimate the effect and the sensitivity of the performance indicators to different policies on the blood supply chain
Detection of subpixel targets with low fill-fraction: An application to hyperspectral imagery
No full textInternational audienceIn hyperspectral imagery, each subpixel target is well-known as the target of interest that only occupies a fraction of the pixel area. The remaining part of the pixel is then filled with the background (at the same spatial location). In this paper, we mainly discuss about a hyperspectral target detector that is represented as a sparse hyperspectral image (HSI) that ideally contains only the subpixel targets with the background is suppressed. More precisely, with the help of a pre-learned target dictionary constructed from some online spectral libraries, the given HSI can be decomposed into a sum of low-rank background HSI and a sparse target HSI, where the latter can be directly used as the target detector. However, with this matrix separation model, the detection of the target of interest may fail (or not succeed without a lot of false alarms) when the subpixel target has a very low fill-fraction and especially when its spectra is well matched to the surroundings. To well alleviate this serious real challenge, we prove via some synthetic experiments, that learning an additional background dictionary and when included in the matrix separation model, would be crucial
Jumeau numérique de processus métier : approche probabiliste augmentée par GNN
No full textNational audienc
Efficient Distance Pruning for process suffix comparison in Prescriptive Process Monitoring
No full textInternational audienceIn an environment where companies seek to optimize the performance of their processes, process mining allows event logs to be analyzed to understand how a process actually runs. Prescriptive Process Monitoring goes further: it aims to recommend the next -best action to improve key performance indicators (KPIs ). However, this type of recommendation relies on comparing a large number of event sequences (called suffixes) to identify those that lead to the best results -a task whose computational cost explodes with the size of the logs. To address this scalability issue, we propose a pruning strategy based on the triangle inequality, where distances to a small set of reference suffixes (pivots) are precomputed to derive bounds and discard unnecessary comparisons without explicit distance calculations
IA pour la recommandation d’actions dans les processus
No full textInternational audienceDans un contexte de pression opérationnelle et d’exigence accrue de performance, les entreprises doivent être capables de réagir rapidement et efficacement. Le process mining permet d’analyser les processus réels à partir des données, tandis que le Prescriptive Process Monitoring vise à recommander la prochaine meilleure action pour améliorer les indicateurs de performance (KPIs). Dans ce cadre, ce travail propose une approche basée sur une architecture JEPA (Joint Embedding Predictive Architecture), combinant des Graph Neural Networks (GNN) pour modéliser les relations entre activités sous forme de multigraphes, et des modèles Time-Aware LSTM pour intégrer les dynamiques temporelles. Cette approche permet de projeter l’état courant d’un processus dans un espace latent afin d’anticiper son évolution et d’évaluer l’impact des actions possibles. Les perspectives incluent l’amélioration de l’explicabilité des recommandations et le renforcement de la prise de décision dans des environnements complexes
Optimizing Saline Buffers for Ultra-Fast and Reliable DNA Hybridization at Room Temperature: A Breakthrough in Nanodiagnostics
No full textInternational audienc
Plasmon-Interband Hybridization and Anomalous Production of Hot Electrons in Aluminum Nanoantennas
No full textInternational audienc
Offline Learning of Maintenance Policies Using Reinforcement Learning and Historical Maintenance Data
No full textInternational audienceIn condition-based maintenance optimization, it is often assumed that the degradation process model is known, so that classical paradigms, such as (Markov)-renewal theory or dynamic programming, can be adopted to find this optimal policy. When degradation modeling becomes challenging, it is possible to learn such a policy directly from maintenance data. Considering offline datasets, consisting of pre-and post-maintenance system states, actions taken and associated costs, generated by various non-optimal behavior policies, our goal is to explore reinforcement learning approach to extract better maintenance policies, without any further system condition monitoring information. In the literature, offline reinforcement learning methods have been studied for maintenance optimization with discounted reward metric and discrete degradation state space, but still received less attention when considering continuous state space in the infinite horizon under the average reward metric. In this paper, we adapted a relative Q-learning algorithm with function approximation to offline settings under the average reward metric and combined it with data augmentation to learn higher performance policies from several maintenance datasets collected from continuously degrading maintained systems. Numerous results under different data configurations show that a nearoptimal policy can be learned with relatively little data
Evaluation de la performance des situations de gestion – le cas des groupes de partenariat opérationnel de la police nationale
No full textInternational audienc
Towards sustainable materials for Nano-optics
No full textInternational audienceFor more than twenty years, noble metal nanoparticles have been of first interest due to their varied and complex optical properties. These properties are mainly governed by collective oscillations of conduction electrons called "plasmons". In particular, the excitation of the plasmon resonance by optical fields leads to a local exaltation of the electromagnetic field close to the nanoparticle. Such very intense nanosource paves the way for numerous applications: controlling, manipulating and amplifying the light at the nanoscale. Nevertheless, if numerous applications have already been developed, the finite stocks of Au and Ag impose us to question some material issues. Is that possible to find abundant materials presenting similar optical properties in order to replace gold and silver? Very recently, new materials have been proposed as suitable materials to tackle emerging applications of nano-photonics, e.g. high-temperature applications, nanochemistry, sensing, or active plasmonics where gold and silver do not possess all the required properties, such as high-temperature sustainability, or catalytic activity. For these reasons, gold and silver are currently reducing their predominance in plasmonics to the benefit of other metals, metal oxides and dielectrics that constitute a new emerging branch of research. Following this burgeoning variety of optical nanomaterials, it becomes useful to conduct a comprehensive and comparative study in order to clearly establish the relative efficiencies of these new materials.In this talk, I will first present all the nanofabrication techniques developed in our group to obtain nanostructures made from materials chosen because of their abundance on earth and optical properties. In this context, I will discuss on the emerging materials and try to present the advantages of different materials like aluminum or silicon