HAL Evry
Not a member yet
19237 research outputs found
Sort by
Investigating general interface effects in one-dimensional phononic crystals for micro-scale sensor application
No full textInternational audienceMicro-scale one-dimensional phononic crystal (PnC) sensors utilizing high-frequency acoustic waves have emerged as a prominent research focus in microfluidic detection. The large surface-to-volume ratios and unique interface properties inherent to micro-composite structures make it imperative to consider general interface effects. In this study, a general interface model is used to systematically investigate the interface effects in 1D PnCs and assess the impact on the performance of micro-scale sensors. The model incorporates interface thickness, stiffness, and inertia to provide a robust, physically realistic representation of interface behavior. Under extreme conditions, the model naturally degenerates into either the spring-layer interface model or the coherent interface model. The results indicate that general interface effects induce significant deviations in defect state PnC resonance frequencies from classical elastic theory predictions. Nevertheless, the resonance frequency remains strongly linearly correlated with the acoustic properties of the measured solution. The performance of 1D PnC sensors improves markedly as their dimensions decrease. In contrast, when the overall structure size is fixed, increases in interface thickness, reductions in stiffness, and higher inertia parameters collectively diminish sensor sensitivity
Distributionally robust master surgery scheduling with duration uncertainty and specialty parallelism
No full textInternational audienceThis study investigates master surgery scheduling at the tactical decision-making level of operating room (OR) management, addressing uncertainty in surgeons’ surgery durations and parallelism in surgical specialties. The goal is to optimize OR time block types within the scheduling cycle, allocate them efficiently to surgical specialties and surgeons, and determine the appropriate number of surgeries to schedule. Given the limited historical data on surgery durations, we employ a distributionally robust optimization (DRO) approach to address the uncertainty in the distribution. To address the needs of different OR managers, we develop a distributionally robust chance-constrained model to manage overtime that extends beyond the designated OR time blocks. Meanwhile, we construct a distributionally robust bi-objective optimization model with the goals of minimizing the expected total duration of overtime and maximizing the number of surgeries scheduled. These optimization models are reformulated into computationally tractable forms using dual theory. We validate the proposed methods with real hospital data, finding that the DRO approach offers greater stability in scheduling solutions compared to the sample average approximation approach
Event-Triggered Controller Design for Multi-Agent Systems
No full textInternational audienceThis chapter presents an event-triggered controller design for multi-agent systems, where the data transmission among agents is governed by event-triggered mechanisms. The dynamic event-triggered control with the full-state observer is introduced, with a particular concern of reducing communication frequency through designable inter-event time. An extension to directed topology, where the communication is restricted to single directions, is also discussed. The proposed methods follow the co-design principle, where the parameters of controllers, observers, and event-triggered mechanisms are synthesized simultaneously. Following a distributed design principle, only local information is required in a small range of nearby neighbors to compute the control signal and the event-triggered condition. Zeno behavior is proved to be excluded. Finally, the proposed approaches are demonstrated by a numerical example and compared with other methods to validate their effectiveness
Digital divide and artificial intelligence for health
No full textFNEGE 2, ABS 3International audienceSocial media platforms have become key intermediaries for ad campaigns, but concerns persist regarding the veracity of information presented in ads. In the health sector, false or unsupported claims in ad content can have real-world public health consequences. On these platforms, the display of ads is managed by recommendation systems that match the content of the ad to the interests of the user. This paper investigates whether the use of AI algorithms to recommend ads on social media platforms may help progress toward the Sustainable Development Goals (SDGs). We collected ads across all US states on Meta and Instagram during a period marked by increased public health concerns. Using a fine-tuned deep learning model, we fact-checked the content of these ads. The results of the fact-check show that only 0.2 % of the ads were classified as misinformation, and 15.41 % of the ads were classified as ambiguous. Both types of ads are less likely to be recommended to users located in wealthier states especially when health-related. Also, health-related ads classified as misinformation are more likely to be recommended to users in states with high percentage of people without health insurance. We argue that the use of recommendation systems contributes to widening the digital divide, which can hinder the achievement of SDGs
Estimator-based event-triggered leader–following consensus of multiagent systems under denial-of-service attacks
No full textInternational audienceThe leader–following consensus issue is investigated for multiagent systems under external disturbances and denial-of-service (DoS) attacks. DoS attacks are considered to be designed according to an unknown attack strategy and would block the communication network. First, a novel triggering mechanism is designed to save communication resources during normal communication periods by event-triggered strategy and to detect the end of attack periods by time-triggered strategy. Then, a set of estimators is constructed to predict the states of the agent itself and its neighbors when the attack occurs. The estimators do not contain the control signals, which only correct the prediction when the actual system states are received. As a result, a distributed switching controller is designed that includes event-triggered states during normal communication periods and predicted states during attack periods. Finally, the proposed secure control protocol is proven to ensure that the system states are eventually consensus and the Zeno behavior is not exhibited. Moreover, a simulation example is given to illustrate the effectiveness of the presented strategy
Crediting football players for creating dangerous actions in an unbiased way: the generation of threat (GoT) indices
No full textInternational audienceWe introduce an innovative methodology to identify football players at the origin of threatening actions in a team. In our framework, a threat is defined as entering the opposing team's danger area. We investigate the timing of threat events and ball touches of players, and capture their correlation using Hawkes processes. Our model-based approach allows us to evaluate a player's ability to create danger both directly and through interactions with teammates. We define a new index, called Generation of Threat (GoT), that measures in an unbiased way the contribution of a player to threat generation. For illustration, we present a detailed analysis of Chelsea's 2016-2017 season, with a standout performance from Eden Hazard. We are able to credit each player for his involvement in danger creation and determine the main circuits leading to threat. In the same spirit, we investigate the danger generation process of Stade Rennais in the 2021-2022 season. Furthermore, we establish a comprehensive ranking of Ligue 1 players based on their generated threat in the 2021-2022 season. Our analysis reveals surprising results, with players such as Jason Berthomier, Moses Simon and Frederic Guilbert among the top performers in the GoT rankings. We also present a ranking of Ligue 1 central defenders in terms of generation of threat and confirm the great performance of some center-back pairs, such as Nayef Aguerd and Warmed Omari
Exploring digital sovereignty through data flows : empirical evidence from the backbone of the internet
No full textInternational audienceThis study investigates the achievability of digital sovereignty within the European Union by examining website data flows, mainly focusing on non-personal data. Amid growing concerns over data governance and the resurgence of digital sovereignty as a central theme in EU policies, this research uniquely addresses firms' data storage decisions and their implications for EU users. Utilizing an original dataset of the most visited websites in France, this paper analyzes data location preferences across various sectors, revealing a complex interplay between privacy regulations, firm size, sector-specific tendencies, and the underlying internet infrastructure. The findings suggest that firms prioritize data storage in countries with strong privacy regulations and tend to locate data closer to consumers to minimize latency. However, variations are observed based on sector-specific needs and firm size, with larger and tech-oriented firms showing less sensitivity to distance. The study also highlights the significant role of the Internet's backbone infrastructure in shaping data storage strategies, pointing to potential challenges in aligning with digital sovereignty goals
Some general external forces and critical mild solutions for the fractional Navier-Stokes equations
No full textIn this article we study mild solutions for the forced, incompressible fractional Navier-Stokes equations. These solutions are classically obtained via a fixed-point argument which relies on suitable estimates for the initial data, the nonlinearity and the external forces. Many functional spaces can be considered, however we are mainly interested here in a critical setting which ensures the existence of global solutions. We give some examples of such critical functional spaces and we discuss their relationship with generic external forces
Biographical Feature: In memoriam Pierre Tiollais (1934–2024)
No full textInternational audiencePierre Tiollais, a French physician-biologist who cloned and sequenced the genome of hepatitis B virus (HBV) and developed a recombinant hepatitis B vaccine, died on 5 August 2024, at the age of 89. Pierre had his wife Irène and his son Romain close by his side through the difficult times since he had been ill. Born on 8 December 1934 in Rennes, Brittany, to pharmacist parents, Pierre remained attached to his native province, which in France is characterized by a strong anthropological particularism, throughout his life and always proudly defined himself as a Breton
Wetting of a dynamically patterned surface is a time-dependent matter
No full textInternational audienceWetting of a dynamically patterned surface is a time-dependent matter In nature and many technological applications, aqueous solutions are in contact with patterned surfaces, which are dynamic over timescales spanning from ps to μs. In biology, exposed polar and apolar residues of biomolecules form a pattern, which fluctu- ates in time due to sidechain and conformational motions. At metal/ and oxide/water interfaces the pattern is formed by surface topmost atoms, and fluctuations are due to, e.g., local surface polarization and rearrangements in the adsorbed water layer. All these dynamics have the potential to influence key processes such as wetting, energy relaxation, and biological function. Yet, their impact on the water H-bond network remains elusive. Here, we leverage on molecular dynamics to address this fundamental question at a Self-Assambled Monolayer (SAM)/water interface, where ns dynamics is induced by frustrating SAM-water interactions via methylation of the terminal -OH groups. We find that surface dynamics couples to the water H-bond network, inducing a response on the same ns timescale. This leads to time fluctuations of local wetting, oscillating from hydrophobic to hydrophilic environments. Our results suggest that more than average properties, it is the local—both in time and space—solvation that determines the chemical-physical properties of dynamically patterned surfaces in water