INRIA a CCSD electronic archive server
Not a member yet
122212 research outputs found
Sort by
Development and assessment of an adhesively bonded assembly geometry adapted for both ELS test and toward hygro-mechanical aging evaluation
No full textInternational audienceStructural adhesive bonding has gained significant attention in recent years. However, its durability in harsh environments, such as marine ones, remains insufficiently examined due to a lack of comprehensive research. One reliable approach for assessing the strength of bonded assemblies is the fracture mechanics, which involves conducting experimental tests in different loading modes to determine their fracture toughness. These tests can be performed before and after aging conditions to evaluate the mechanical properties evolution of the bonded assembly in response to aging. While mode I may occur, mode II is usually dominant in bonded joints; thus, only mode II is considered in this study. The selected mode II test is the End-Loaded Split (ELS) test. Most studies have focused on either the mechanical test setup or the aging setup, but have lacked a combined approach. This paper introduces a new sample configuration that merges both the ELS test setup and the hygro-mechanical frame. To validate the new sample design, a comparison of the fracture toughness values and of the cohesive zone models (CZMs), obtained after fracture data elaboration, is made between this new configuration and a standard ELS sample geometry. For the fracture toughness measurement, all specimens were instrumented with distributed optical fiber sensor to monitor crack propagation as well as Digital Image Correlation system to calculate the J-integral values from which the CZM can be derived. One major conclusion is that the results are repeatable, confirming that the new sample design is valid for the upcoming aging study
Physics-Informed Neural Estimation of State and Unknown Input inAutonomic Cardiac Dynamics with Left-Invertibility Constraints
No full textInternational audienceUnderstanding brain–heart interaction (BHI) requires models that capture how the central nervous system and the cardiovascular system co-regulate each other under internal and external stressors to preserve homeostasis and give rise to macroscopic physiological states such as sleep, arousal, or vigilance. At the core of this loop are interoceptive variables, defined as latent autonomic control signals that encode the body’s internal state and drive cardiac adjustments; however, these variables are not directly measurable. Recovering these hidden drives from peripheral cardiac signals requires taking into account nonlinear dynamics and physiological confounds, as well as limited measurement data. This work proposes a physics-informed neural estimator for simultaneous state estimation and reconstruction of unknown control inputs in a nonlinear, coupled model of autonomic cardiac regulation. The estimator enforces model-based constraints along with data-driven regularization, and embeds structural identifiability conditions derived from the system itself, yielding guarantees without prescribing a dynamical prior for the unknown input. Validation of stressevokedcardiac recordings shows the accurate recovery of heart rate, along with estimates of state (being blood pressure) and the unknown control input (being the blood-pressure setpoint); hence, enabling physiology-consistent, left-invertible inferenceof interoceptive autonomic dynamics
KAN-SAs: Efficient Acceleration of Kolmogorov-Arnold Networks on Systolic Arrays
No full textInternational audienceKolmogorov-Arnold Networks (KANs) have garnered significant attention for their promise of improved parameter efficiency and explainability compared to traditional Deep Neural Networks (DNNs). KANs' key innovation lies in the use of learnable non-linear activation functions, which are parametrized as splines. Splines are expressed as a linear combination of basis functions (B-splines). B-splines prove particularly challenging to accelerate due to their recursive definition. Systolic Array (SA)based architectures have shown great promise as DNN accelerators thanks to their energy efficiency and low latency. However, their suitability and efficiency in accelerating KANs have never been assessed. Thus, in this work, we explore the use of SA architecture to accelerate the KAN inference. We show that, while SAs can be used to accelerate part of the KAN inference, their utilization can be reduced to 30%. Hence, we propose KAN-SAs, a novel SA-based accelerator that leverages intrinsic properties of B-splines to enable efficient KAN inference. By including a nonrecursive B-spline implementation and leveraging the intrinsic KAN sparsity, KAN-SAs enhances conventional SAs, enabling efficient KAN inference, in addition to conventional DNNs. KAN-SAs achieves up to 100% SA utilization and up to 50% clock cycles reduction compared to conventional SAs of equivalent area, as shown by hardware synthesis results on a 28nm FD-SOI technology. We also evaluate different configurations of the accelerator on various KAN applications, confirming the improved efficiency of KAN inference provided by KAN-SAs
Robust Detection of Synthetic Tabular Data under Schema Variability
No full textInternational audienceThe rise of powerful generative models has sparked concerns over data authenticity. While detection methods have been extensively developed for images and text, the case of tabular data, despite its ubiquity, has been largely overlooked. Yet, detecting synthetic tabular data is especially challenging due to its heterogeneous structure and unseen formats at test time. We address the underexplored task of detecting synthetic tabular data "in the wild'', i.e. when the detector is deployed on tables with variable and previously unseen schemas. We introduce a novel datum-wise transformer architecture that significantly outperforms the only previously published baseline, improving both AUC and accuracy by 7 points. By incorporating a table-adaptation component, our model gains an additional 7 accuracy points, demonstrating enhanced robustness. This work provides the first strong evidence that detecting synthetic tabular data in real-world conditions is feasible, and demonstrates substantial improvements over previous approaches. Following acceptance of the paper, we are finalizing the administrative and licensing procedures necessary for releasing the source code. This extended version will be updated as soon as the release is complete
Infinitely many saturated travelling waves for a degenerate Fisher-KPP equation not in divergence form
No full textInternational audienceWe consider an epidemic model with distributed-contacts. When the contact kernel concentrates, one formally reaches a very degenerate Fisher-KPP equation with a diffusion term that is not in divergence form. We make an exhaustive study of its travelling waves. For every admissible speed, there exist not only a unique non-saturated (smooth) wave but also infinitely many saturated (sharp) ones. Furthermore their tails may differ from what is usually expected. These results are thus in sharp contrast with their counterparts on related models
What can we do in a symmetry-constrained perspective? The importance of the total charge's status in quantum reference frame frameworks
No full text25+6 pages. Comments are welcome!The study of quantum reference frames has received renewed interest over the lastyears, leading to the parallel development of non-equivalent frameworks by different com-munities. We clarify the differences between these frameworks. At the mathematical level,they mainly differ in the kind of symmetry (either weak or strong) employed to constrainthe system. We show that this mathematical difference corresponds to a fundamentalphysical question: whether the global charge associated to the symmetry group is acces-sible to symmetry-constrained observers. In this context, we formulate a definition of aperspective in terms of operational capacities, or lack thereof. Turning to consequences ofadopting either approach, we discuss how adopting the weak approach induces an ambi-guity in the momenta included in each perspective and bars from defining reversible QRFtransformations. We then review and analyze the existing arguments motivating eachapproach, and show how they bear upon the problem of charge accessibility. Finally, weintroduce a simple operational scenario in which upholding two reasonable physical pos-tulates leads to the conclusion that internal observers could measure the global charge by1/ performing a relativized interference measurement and 2/ classically communicating
Neoadjuvant immune checkpoint blockade before chemoradiation for cervical squamous carcinoma (GINECO window-of-opportunity COLIBRI study): a phase II trial
No full textInternational audienc
libMeshb: a Simple, Fast and Versatile Library to Handle HPC Meshes and Solutions with a Dedicated File Format
No full textInternational audienceProblem statement: a full HPC toolchain is made of numerous pre/post-processing tools and solvers, all communicating with each other either through file-based I/O or in memory exchanges. The first solution is less invasive in terms of software development but requires lots of external libraries, components and knowledge to use them (i.e. MPIO, HDF5, GPFS, Lustre, CGNS). The latter is more invasive as internal data structures need to be adapted and exposed to the communication layer but is much faster both in terms of throughput and latency. Furthermore, most standard HPC I/O libraries and file formats have been developed in the time of spinning hard drives and fail to extract the full potential of today's SSDs.Solution: we propose a lightweight library made of a single C file, without external dependency aside the Linux kernel's async IO and pthreads, whose API is based on tables and scalars so it can be easily accessed from any kind of languages and that is inherently parallel and SSD-native
