HAL Paris Dauphine-PSL
Not a member yet
21932 research outputs found
Sort by
Image = Cartoon+Texture: How Yves Meyer's "Oscillating patterns in image processing and in some nonlinear evolution equations" ended up in a computer vision model
A. Bonami, S. Jaffard, S. Seuret, Editeurs.International audienceWhile analyzing variational restoration methods in imaging with the eye of an harmonic analyst, Yves Meyer proposed a seminal reformulation of image decomposition methods, which aim at separating the "cartoon" and "texture" parts of an image. In this note, we describe Meyer's analysis of the image denoising problem. This analysis occupied an important part of his book "Oscillating patterns in image processing and in some nonlinear evolution equations". Starting from the famous Rudin, Osher and Fatemi (ROF) Total Variationbased model for natural image denoising, Meyer tries to bracket the space BV , of functions with bounded variation, and its dual between spaces that can be characterized by the decay of their wavelet coefficients. In that way, he first attempts to build a bridge between the Donoho and Johnstone wavelet shrinkage model for image denoising and the ROF model. This discussion leads him to explore the consistency of the ROF model on examples. He proves that the ROF model sometimes finds the right decomposition, namely BV + an oscillatory component, and sometimes not. This leads him to pay attention to the dual of BV and eventually to the discovery that small norms in this space characterize oscillatory signals very well. This exploration yields a reinterpretation and extension of the ROF model, with the proposal of decomposing a distribution f = u + v into two components, the first one u in BV and another one in a dual space where oscillating components have a small norm, in the spirit of interpolation theory. We illustrate the resulting cartoon+texture model and its developments with striking experiments which show its impact on medicine and society
Le travail déplacé : plaidoyer pour une démocratie managériale sans lieu
International audienceCes dernières années ont été jalonnées de points de rupture pour l’organisation du travail et son management. La crise du sens au travail, les limites planétaires, la pandémie, la guerre, ont radicalement questionné notre relation au travail. Un dé-placement et une démocratisation sont nécessaires
Why are inner planets not inclined?
International audiencePoincaré’s work more than one century ago, or Laskar’s numerical simulations from the 1990’s on, have irrevocably impaired the long-held belief that the Solar System should be stable. But mathematical mechanisms explaining this instability have remained mysterious. In 1968, Arnold conjectured the existence of “Arnold diffusion” in celestial mechanics. We prove Arnold’s conjecture in the planetary spatial 4-body problem as well as in the corresponding hierarchical problem (where the bodies are increasingly separated), and show that this diffusion leads, on a long time interval, to some large-scale instability. Along the diffusive orbits, the mutual inclination of the two inner planets is close to π/2, which hints at why even marginal stability in planetary systems may exist only when inner planets are not inclined. More precisely, consider the normalised angular momentum of the second planet, obtained by rescaling the angular momentum by the square root of its semimajor axis and by an adequate mass factor (its direction and norm give the plane of revolution and the eccentricity of the second planet). It is a vector of the unit 3-ball. We show that any finite sequence in this ball may be realised, up to an arbitrary precision, as a sequence of values of the normalised angular momentum in the 4-body problem. For example, the second planet may flip from prograde nearly horizontal revolutions to retrograde ones. As a consequence of the proof, the non-recurrent set of any finite-order secular normal form accumulates on circular motions – a weak form of a celebrated conjecture of Herman
Classifying bulk-edge anomalies in the Dirac Hamiltonian
International audienceWe study the Dirac Hamiltonian in dimension two with a mass term and a large momentum regularization, and show that bulk-edge correspondence fails. Despite a well defined bulk topological index --the Chern number--, the number of edge modes depends on the boundary condition. The origin of this anomaly is rooted in the unbounded nature of the spectrum. It is detected with Levinson's theorem from scattering theory and quantified via an anomalous winding number at infinite energy, dubbed ghost charge. First we classify, up to equivalence, all self-adjoint boundary conditions, using Schubert cell decomposition of a Grassmanian. Then, we investigate which ones are anomalous. We expand the scattering amplitude near infinite energy, for which a dominant scale captures the asymptotic winding number. Remarkably, this can be achieved for every self-adjoint boundary condition, leading to an exhaustive anomaly classification. It shows that anomalies are ubiquitous and stable. Boundary condition with a ghost charge of 2 is also revealed within the process
Made in China luxury: why Chinese generations matter
Luxury: Story of Heritage, Innovation and Disruption in the AI EraSince our last symposium two years ago, we have witnessed uncertainty and disruption in luxurymarkets. The personal luxury goods sector experienced its first decline in 15 years (excluding the COVIDperiod), the luxury goods consumer base contracted by 50 million people and further deepening thesechallenges, sales and margins have been decreasing for luxury brands. And yet, here we are, our presence testament to the unwavering confidence we all have in the resilience of luxury and importance of luxuryresearch. In spite of the complexity of these challenges, the luxury industry is expected to undergo one of itsmost remarkable transformations, influenced by new technologies, changing consumer needs, and greaterdemands for sustainable and responsible business practices. We look forward to discussing these challenges, responses and outcomes with you in the context of: “Luxury: a Story of Heritage, Innovation and Disruption in the AI Era”Since late last year we have had the pleasure of receiving your research papers. We are honored that youhave chosen to share your work with us and deeply appreciate the rigor and originality of your contributions. Your research is invaluable to the success of this symposium, and we sincerely thank you. The symposium would not have been possible without the support of the many reviewers who gave their time and expertise to thoughtfully reviewing these submissions. Thank you for your generous contribution. Finally, we would like to thank everyone who was involved in the successful organization of this event, without their contribution this symposium would not be possible: Jean-Philippe Muller (General Director & Dean of IUM), Sophie de Lorenzo (Director Corporate Relations, IUM), Iulia Pantea (Events & Communication Manager, IUM), , Mariateresa Torchia (Professor of Management, Head of Research & DBA Program Director, IUM), Anne Marie Vesdrevanis (Professor of Management, IUM), Carolina Bracco (Communication & Recruitment Manager, IUM) and OMNES Education. On that note, we are honored to present you with the 2025 Monaco Symposium of Luxury research abstracts. We hope that you enjoy reading them as much as we did and we look forward to an insightful and collaborative symposium
Small Trees Take Down Big Axes
International audienceAs part of a discussion of the author’s Une écologie décoloniale: Penser l’écologie depuis le monde caribéen (2019), translated by Anthony Paul Smith as Decolonial Ecology: Thinking from the Caribbean World (2022), this essay is in response to the essays by Alex A. Moulton and Sophie Large. First, the author more precisely articulates ecology with decolonial emancipation demands, including the need for a reinvention of our relationships with other-than-human animals. Second, he argues for the need to decenter Whiteness and coloniality in the appraisal of conceptual works that do not follow the focus of canonical European theory or the South American roots of decolonial theory. In conclusion, “small trees breaking down big axe” echoes the core proposition of Decolonial Ecology—that is, a political theory and aesthetics that holds together the need for the ecological preservation of Earth and the imperative of decolonial emancipation in the wake of euromodern colonization and slavery
Forward Only Learning for Orthogonal Neural Networks of Any Depth
International audienceBackpropagation is still the de facto algorithm used today to train neural networks. With the exponential growth of recent architectures, the computational cost of this algorithm also becomes a burden. The recent PEPITA and forward-only frameworks have proposed promising alternatives, but they failed to scale up to a handful of hidden layers, yet limiting their use. In this paper, we first analyze theoretically the main limitations of these approaches. It allows us the design of a forward-only algorithm, which is equivalent to backpropagation under the linear and orthogonal assumptions. By relaxing the linear assumption, we then introduce FOTON (Forward-Only Training of Orthogonal Networks) that bridges the gap with the backpropagation algorithm. Experimental results show that it outperforms PEPITA, enabling us to train neural networks of any depth, without the need for a backward pass. Moreover its performance on convolutional networks clearly opens up avenues for its application to more advanced architectures. The code is open-sourced on https://github.com/p0lcAi/FOTON
Resurrecting the Dead: GenAI-Powered Workflow Revival
International audienceWhile scientific workflows have been established and used in a number of disciplines for specifying and executing experiments and data analysis, early and recent studies have demonstrated that an important proportion of workflows suffer from decay. This phenomena is exacerbated by legacy scientific workflow systems, notably Taverna, which was popular in e-science for orchestrating complex analyses.A step towards addressing this issue, we report on in this paper a feasibility study on using generative AI to revive decayed workflows, combining large language models with modern workflow technologies. Our approach automates critical revival tasks including parsing of legacy Taverna workflows, failure point identification, repair suggestion, and conversion to contemporary formats, viz. SnakeMake. The methodology integrates AI-driven workflow summarization, pseudocode abstraction, graph-based visualization, automated service substitution, and code generation. We demonstrate and evaluate this approach through a real-world decayed workflow case study. We conclude the paper with a discussion on key lessons that we learned and will guide development of a systematic workflow revival framework as part of our future work
Pénuries d'eau : innovations technologiques et nouveaux modèles économiques
Simon PORCHER, Professeur à l’Université Paris Dauphine-PSL et auteur de "La fin de l'eau?" Editions Fayard, était notre invitée à cette 4ème séance du cycle 2024-2025 des séminaires du Cercle de l'innovation. Face à la raréfaction de l'eau, la gestion de cette ressource impose de repenser l'offre, la demande et le modèle économique associé. Stabiliser l'offre grâce à des technologies comme le retraitement des eaux usées, la désalinisation ou le captage de l'eau des brouillards, est nécessaire dans certains territoires mais ne doit pas encourager une surconsommation. L'enjeu réside aussi dans la valorisation de la sobriété permettant une meilleure connaissance et préservation de la ressource. Un modèle économique favorisant l'usage raisonné plutôt que le gaspillage devient essentiel pour faire face aux défis à venir