HAL-INSA Toulouse
Not a member yet
34325 research outputs found
Sort by
Fairness is in the details: Face Dataset Auditing
No full textAuditing involves verifying the proper implementation of a given policy. As such, auditing is essential for ensuring compliance with the principles of fairness, equity, and transparency mandated by the European Union's AI Act. Moreover, biases present during the training phase of a learning system can persist in the modeling process and result in discrimination against certain subgroups of individuals when the model is deployed in production. Assessing bias in image datasets is a particularly complex task, as it first requires a feature extraction step, then to consider the extraction's quality in the statistical tests. This paper proposes a robust methodology for auditing image datasets based on so-called "sensitive" features, such as gender, age, and ethnicity. The proposed methodology consists of both a feature extraction phase and a statistical analysis phase. The first phase introduces a novel convolutional neural network (CNN) architecture specifically designed for extracting sensitive features with a limited number of manual annotations. The second phase compares the distributions of sensitive features across subgroups using a novel statistical test that accounts for the imprecision of the feature extraction model. Our pipeline constitutes a comprehensive and fully automated methodology for dataset auditing. We illustrate our approach using two manually annotated datasets
NAPALAIR, une recherche interdisciplinaire et participative autour d'un échange écologique inégal
No full textInternational audienc
A Whole-Body Multi Contact Large Object Manipulation and Estimation Framework for Humanoids Using Skin Patches
No full textInternational audienceOver the years, robotic manipulation has primarily focused on end-effectors, an approach that severely limits a robot's ability to manipulate large and heavy objects. Humanoids, which are expected to operate in human environments, must acquire this skill to enhance their versatility and usefulness. In this regard, we present a whole-body multicontact manipulation (WBMC) framework to handle large and heavy objects. To facilitate WBMC, we incorporate artificial skin patches distributed across the humanoid's upper body which are effectively utilized for contact detection and force sensing. The WBMC manipulation problem is formulated as an optimal control problem (OCP) within a model predictive control (MPC) framework, and three different types of dynamic motions are used to evaluate the controller's effectiveness. The proposed framework manages the entire manipulation process, including reaching, grasping, picking up, and manipulating. Furthermore, these motions are leveraged to develop a twostage object inertial parameter estimation framework. The first stage estimates the object's mass and center of mass, while the second estimates its inertia. Both the manipulation and estimation frameworks are numerically evaluated using the TALOS humanoid and a rectangular box in simulation, and their respective results are presented and discussed
Modélisation des Systèmes Dynamiques pour l'analyse d'alternatives plus circulaires de gestion des déchets dans les centres commerciaux
No full textInternational audienc
Automatic visual inspection of mechanical assemblies via 3D point cloud classification with deep neural networks
No full textIn this work, we are focused on conformity control of complex aeronautical mechanical assemblies,typically an aircraft engine, at the end or in the middle of the assembly process. A 3D scannercarried by a robot arm provides acquisitions of 3D point clouds which are further processed by deepclassification neural networks. The Computer-Aided Design (CAD) model of the inspected mechanicalassembly is available, and the proposed approach relies on it. The models are trained on syntheticdata, generated from the CAD models
Consequential Life Cycle Assessment of Bio-Phenol Recovery from Softwood Pyrolysis
No full textInternational audienceThe significant utilisation of renewable biomass remains difficult primarily because of their complex biochemical constituents and low selectivity for its end products. Herein, we recommend a biomass pyrolysis followed by a solvent extraction process to efficiently recover phenol while also generating valuable co-products that can substitute fossil-based resources and reduce overall fossil energy demand. A detailed consequential life cycle assessment (LCA) was performed to quantify the energy use and environmental impacts of the pyrolysis biorefinery as compared to traditional approaches. Additionally, the use of pyrolysed biochar as a soil amendment application contributes to carbon negativity, while the combustion of bio-oil and non-condensable gases and wood dust generates heat and electricity, partially substituting conventional energy sources. The result highlights biomass supply chains such as indirect land use change, type of pyrolysis technology, yield of co-product and its application, also choice of marginal suppliers and these factors have a direct impact on the environmental performance of pyrolysis based biorefineries. Also, the sensitivity analysis one at the time (OAT) of the major parameters identifies possibilities for process optimisation, such as reducing the energy consumption and increasing phenol production. Results obtained in this study are intended to deliver the evidence for industrial stakeholders and policymakers to make the right decisions and support investments toward a low fossil carbon future
Orbital Stability of Plane Waves in the Klein-Gordon Equation against Localized Perturbations
No full textWe investigate the stability and long-term behavior of spatially periodic plane waves in the complex Klein-Gordon equation under localized perturbations. Such perturbations render the wave neither localized nor periodic, placing its stability analysis outside the scope of the classical orbital stability theory for Hamiltonian systems developed by Grillakis, Shatah, and Strauss. Inspired by Zhidkov's work on the stability of time-periodic, spatially homogeneous states in the nonlinear Schrödinger equation, we develop an alternative method that relies on an amplitude-phase decomposition and leverages conserved quantities tailored to the perturbation equation. We establish an orbital stability result of plane waves that is locally uniform in space, accommodating L2-localized perturbations as well as nonlocalized phase modulations. In certain regimes, our method even allows for unbounded modulations. Our result is sharp in the sense that it holds up to the spectral stability boundary
Trend to equilibrium and diffusion limit for the inertial Kuramoto-Sakaguchi equation
No full textIn this paper, we study the inertial Kuramoto-Sakaguchi equation for interacting oscillatory systems. On the one hand, we prove the convergence toward corresponding phase-homogeneous stationary states in weighted Lebesgue norm sense when the coupling strength is small enough. In [10], it is proved that when the noise intensity is sufficiently large, equilibrium of the inertial Kuramoto-Sakaguchi equation is asymptotically stable. For generic initial data, every solutions converges to equilibrium in weighted Sobolev norm sense. We improve this previous result by showing the convergence for a larger class of functions and by providing a simpler proof. On the other hand, we investigate the diffusion limit when all oscillators are identical. In [19], authors studied the same problem using an energy estimate on renormalized solutions and a compactness method, through which error estimates could not be discussed. Here we provide error estimates for the diffusion limit with respect to the mass m ≪ 1 using a simple proof by imposing slightly more regularity on the solution
A Complex Network Analysis Approach for Generating Realistic Instances of the Scheduled Service Network Design Problem
No full textLong-haul freight transportation forms the backbone of global supply chains and involves diverse types of carriers, e.g., liner shipping companies, rail freight operators, less-than-truckload carriers, express parcel companies, etc. Planning decisions across these applications can be assisted by solving the Scheduled Service Network Design Problem, a fundamental but computationally challenging optimization problem. However, most instances used in the literature are not accessible, and the few available benchmarks were not designed to reflect the structural properties of real-world freight transportation networks. To fill this gap, we introduce a new open-source generator of instances for the Scheduled Service Network Design Problem. Our generator leverages insights from Complex Network Analysis to reproduce key structural features of freight transportation networks, while allowing users to tune parameters to generate diversified instances. A computational study validates its ability to produce networks exhibiting metrics aligned with those of networks derived from existing freight applications. We also include features such as preprocessing rules, varied demand pattern generation, and network emulation to increase the value of our tool for both practitioners and researchers
Multivariate Bernoulli Hoeffding Decomposition: From Theory to Sensitivity Analysis
No full textUnderstanding the behavior of predictive models with random inputs can be achieved through functional decompositions into sub-models that capture interpretable effects of input groups. Building on recent advances in uncertainty quantification, the existence and uniqueness of a generalized Hoeffding decomposition have been established for correlated input variables, using oblique projections onto suitable functional subspaces. This work focuses on the case of Bernoulli inputs and provides a complete analytical characterization of the decomposition. We show that, in this discrete setting, the associated subspaces are one-dimensional and that the decomposition admits a closed-form representation. One of the main contributions of this study is to generalize the classical Fourier–Walsh–Hadamard decomposition for pseudo-Boolean functions to the correlated case, yielding an oblique version when the underlying distribution is not a product measure, and recovering the standard orthogonal form when independence holds. This explicit structure offers a fully interpretable framework, clarifying the contribution of each input combination and theoretically enabling model reverse engineering. From this formulation, explicit sensitivity measures—such as Sobol’ indices and Shapley effects—can be directly derived. Numerical experiments illustrate the practical interest of the approach for decision-support problems involving binary features. The paper concludes with perspectives on extending the methodology to high-dimensional settings and to models involving inputs with finite, non-binary support.</div