HAL Portal IOGS (nstitut d'Optique Graduate School)
Not a member yet
12589 research outputs found
Sort by
Vibration analysis of transversely loaded arches using Hencky bar-chain model
No full textInternational audienc
Augmented Vision Systems: Paradigms and Applications
No full textInternational audienceAugmented Reality (AR) has grown from specialised uses to applications for the common public. One of these developments led to Augmented Vision (AV), which enhances vision beyond traditional methods like glasses or contact lenses. This review aims to compare and categorise AV systems according to the paradigms they implement to enhance the users' vision.Additionally, the review examines whether researchers conduct measurements and analysis on the human visual system (HVS) when evaluating their system. Such an overall view will help future researchers position their work on AV. By understanding AV systems' paradigms and approaches, researchers will be well-equipped to identify gaps, explore novel directions, and leverage existing advancements.We searched Scopus, Web of Science, and PubMed databases for publications until February 26, 2025, exploring citations and references for the selected articles to avoid missing out on relevant articles. We then conducted a two-step screening process that involved LLM-assisted screening of the article's abstracts and an in-depth assessment of the article. This review follows the PRISMA statement, reducing bias risk. We selected 113 of 469 articles, as they improved users' visual performance. We defined three main categories: (1) adding light to the incoming light field, (2) modifying the incoming light field, and (3) intersecting approaches. We found three main application areas: (1) task-specific, (2) vision correction, and (3) visual perception enhancement. The most typical application is task-specific. We identified a gap in the literature since just four of the papers we reviewed measured and analysed the accommodation while utilising the device
Explications contrefactuelles efficaces et effectives pour les forêts aléatoires
No full textInternational audienceRandom forests are widely used in machine learning due to their excellent predictive performance and computational efficiency. However, their inherent complexity often hinders interpretability, making it challenging for users to understand the decision-making process. Explainable Artificial Intelligence (XAI) techniques aim to mitigate this issue by improving model transparency and providing explanations for predictions. Among these techniques, counterfactual explanations provide intuitive insights by describing the minimal modifications needed to achieve a desired outcome. However, generating counterfactual explanations of good quality is still challenging for random forests. In this paper, we propose a novel explanation approach called Efficient and Effective Counterfactual Explanation (EECE) for random forests, which generates counterfactual explanations by leveraging the structure of decision tree leaves. EECE not only ensures efficient explanation generation but also satisfies essential properties for high-quality counterfactual explanations, such as validity, proximity, sparsity, diversity, plausibility, and actionability. We compare EECE with existing methods across 15 datasets using multiple evaluation metrics, demonstrating its effectiveness in generating high-quality counterfactual explanations.</div
Provably Accurate Adaptive Sampling for Collocation Points in Physics-informed Neural Networks
No full text20 pages.International audienceDespite considerable scientific advances in numerical simulation, efficiently solving PDEs remains a complex and often expensive problem. Physics-informed Neural Networks (PINN) have emerged as an efficient way to learn surrogate solvers by embedding the PDE in the loss function and minimizing its residuals using automatic differentiation at so-called collocation points. Originally uniformly sampled, the choice of the latter has been the subject of recent advances leading to adaptive sampling refinements for PINNs. In this paper, leveraging a new quadrature method for approximating definite integrals, we introduce a provably accurate sampling method for collocation points based on the Hessian of the PDE residuals. Comparative experiments conducted on a set of 1D and 2D PDEs demonstrate the benefits of our method
Homogeneous magnetic flux in Rydberg lattices
No full textInternational audienceWe present a method for generating homogeneous and tunable magnetic flux for bosonic particles in a lattice using Rydberg atoms. Our setup relies on Rydberg excitations hopping through the lattice by dipolar exchange interactions. The magnetic flux arises from complex hopping via ancilla atoms. Remarkably, the total flux within a magnetic unit cell directly depends on the ratio of the number of lattice sites to ancilla atoms, making it topologically protected to small changes in the positions of the atoms. This allows us to optimize the positions of the ancilla atoms to make the flux through the magnetic unit cell homogeneous. With this homogeneous flux, we get a topological band in the single-particle regime. In the many-body regime, we obtain indications of a bosonic fractional Chern insulator state at ν = 1/2 filling
Spin-1 Haldane Phase in a Chain of Rydberg Atoms
No full textInternational audienceWe present a protocol to implement a spin-1 chain in Rydberg systems using three Rydberg states close to a Förster resonance. In addition to dipole-dipole interactions, strong van der Waals interactions naturally appear due to the presence of the Förster resonance and give rise to a highly tunable Hamiltonian. The resulting phase diagram is studied using the infinite density-matrix renormalization group and reveals a highly robust Haldane phase-a prime example of a symmetry-protected topological phase. We find experimentally accessible parameters to probe the Haldane phase in current Rydberg systems, and demonstrate an efficient adiabatic preparation scheme. This paves the way to probe the remarkable properties of spin fractionalization in the Haldane phase.</div
Brain photobiomodulation: a potential treatment in Alzheimer’s and Parkinson’s diseases
No full textInternational audienceAlzheimer's Disease (AD) and Parkinson's Disease (PD) are common neurodegenerative diseases, characterized by the progressive loss of synapses and neurons, leading to cognitive and motor decline. Their pathophysiology includes cerebral lesions, oxidative stress, neuroinflammation as well as brain-gut axis microbiota dysbiosis. Preclinical investigations demonstrated that brain photobiomodulation (bPBM) reduces oxidative stress and inflammation, increases cerebral blood flow and enhance neurogenesis and synaptogenesis, which makes bPBM a promising treatment in AD and PD. This review focuses on the clinical application of bPBM in AD and PD. It aims to provide a scientific overview of the current clinical knowledge, review recent clinical studies findings, and describe future directions and upcoming clinical studies. So far, several clinical studies investigated bPBM therapy, at various parameters, both in patients with AD and related dementia, and PD. All demonstrate bPBM safety and bring valuable clinical information regarding efficacy, with particularly promising results in AD. However, their exploratory design and inconsistent quality lead to a low level of evidence, which currently does not support the widespread use of bPBM in clinical practice. Future clinical research should address two gaps: the need for robust double-blinded RCTs vs sham with a higher number of patients and a longer follow-up, and the need for research focusing on dosimetry to determine which bPBM parameters are optimal. The ongoing or unpublished clinical studies on bPBM should fill in this gap.</div
Pulsed laser engineering of composite submicron particles in colloidal systems: a high-performance catalyst for ethanol fuel cells
No full textInternational audienceNanoparticles are widely regarded as optimal for catalytic reactions; however, larger particles with highly active surfaces may offer an intriguing alternative for advancing catalytic technologies. This study employs pulsed laser melting to transform colloidal copper/magnetite nanoparticles into surface-active submicron CuxFe3-xO4-CuyO-CuzFe1-z composite particles, tailored for ethanol oxidation fuel cells. The findings reveal that colloidal particles tend to cluster into either homogeneous or heterogeneous aggregates, mediated by the surrounding liquid. This clustering aids the formation of desired phases during pulsed laser processing. Temperature-dependent thermodynamic phase transitions, combined with pulse-driven heating-cooling dynamics, promote copper oxidation and magnetite reduction, achieving both compositional control and microstructural surface activation. The synthesized heterostructures demonstrated excellent performance in ethanol oxidation, both as primary catalytic materials and as activity-enhancing supports for platinum. Oxidation state analysis post electrocatalysis indicated a reduction in graphite bonds and an increase in oxygen bonds, attributed to the high oxygen content of the catalysts’ surface. The electrocatalysis ethanol oxidation process generated potent oxidizing agents, including ozone, oxygen and hydroxyl radicals, with the ability of degrading the sp2 hybrid structure of graphite. Despite their submicron size, the kinetically activated composite particles exhibited exceptional surface activity, positioning them as cost-effective alternatives to the conventional catalysts for fuel cell technologies
Photonic Self-Learning in Ultrafast Laser-Induced Complexity
No full textHow can one design complex systems capable of learning for a given functionality? In the context of ultrafast laser-surface interaction, we unravel the nature of learning schemes tied to the emergence of complexity in dissipative structures. The progressive development of learning mechanisms, from direct information storage to the development of smart surfaces, originates from the network of curvatures formed in the unstable fluid under thermoconvective instability, which is subsequently quenched and resolidified. Under pulsed laser irradiation, non-equilibrium dynamics generate intricate nanoscale patterns, unveiling adaptive process mechanisms. We demonstrate that the imprints left by light act as a form of structural memory, encoding not only local effects directed by laser field polarization but also a cooperative strategy of reliefs that dynamically adjust surface morphology to optimize light capture. By investigating how apparent complexity and optical response are intricately intertwined, shaping one another, we establish a framework that draws parallels between material adaptation and learning dynamics observed in biological systems