INRIA a CCSD electronic archive server
Not a member yet
122212 research outputs found
Sort by
Impact of a moving-axis knee model on musculoskeletal forces during gait
No full textInternational audienc
Human-Robot Co-Simulation method for upper limb assistive force calculation using polytopes
No full textInternational audienc
Quasi-compactness for dominated kernels with application to quasi-stationary distribution theory
No full textWe establish a domination principle for positive operators, yielding a upper bound on the essential spectral radius and a practical quasi-compactness criterion on weighted supremum spaces. We then apply these results to absorbed Markov processes and show that quasi-compactness of the transition kernel ensures existence and convergence to quasi-stationary distributions in broadly reducible settings, without regularity requirements. In continuous time, we show that measurability plus quasi-compactness at a single time propagates to all times, rules out periodic behavior, and yields convergence to quasi-stationary distributions. Two illustrative cases demonstrate the scope and simplicity of the criteria.</div
AtomSurf : Surface Representation for Learning on Protein Structures
No full textInternational audienceWhile there has been significant progress in evaluating and comparing different representations for learning on protein data, the role of surface-based learning approaches remains not well-understood. In particular, there is a lack of direct and fair benchmark comparison between the best available surface-based learning methods against alternative representations such as graphs. Moreover, the few existing surface-based approaches either use surface information in isolation or, at best, perform global pooling between surface and graph-based architectures. In this work, we fill this gap by first adapting a state-of-the-art surface encoder for protein learning tasks. We then perform a direct and fair comparison of the resulting method against alternative approaches within the Atom3D benchmark, highlighting the limitations of pure surface-based learning. Finally, we propose an integrated approach, which allows learned feature sharing between graphs and surface representations on the level of nodes and vertices across all layers. We demonstrate that the resulting architecture achieves state-of-the-art results on all tasks in the Atom3D benchmark, while adhering to the strict benchmark protocol, as well as more broadly on binding site identification and binding pocket classification. Furthermore, we use coarsened surfaces and optimize our approach for efficiency, making our tool competitive in training and inference time with existing techniques. Code can be found online: https://github.com/Vincentx15/atomsur
Stable Extrapolation of Field Produced by Volumetric Magnetizations
No full textInternational audienceIn the particular setup of the planar scanning Superconducting Quantum Interference Device (SQUID), the vertical component of the magnetic field produced by a magnetized sample is measured. Recovering the sample’s internal magnetization from this measured data, an inverse problem that is central in the field of paleomagnetism, is a severely ill-posed process. Moreover, standard recovery methods are further hindered by limited measurements and the noise therein. To address these issues, we develop a method to simultaneously extrapolate and denoise the field data, thereby solving a preliminary inverse problem for an auxiliary function of the magnetization. The proposed approach is based on a regularization framework that exploits an explicit field-magnetization relation. To encode the harmonic structure of the problem, we construct a set of basis functions derived from spherical harmonics via the Kelvin transform. The method is applicable to both volumetric and planar magnetization distributions
A new bound on the cumulant generating function of Dirichlet processes
No full textInternational audienceIn this paper, we introduce a novel approach for bounding the cumulant generating function (CGF) of a Dirichlet process (DP) , using superadditivity. In particular, our key technical contribution is the demonstration of the superadditivity of , where . This result, combined with Fekete's lemma and Varadhan's integral lemma, converts the known asymptotic large deviation principle into a practical upper bound on the CGF for any α> 0. The bound is given by the convex conjugate of the scaled reversed Kullback-Leibler divergence . This new bound provides particularly effective confidence regions for sums of independent DPs, making it applicable across various fields
Fault-tolerant transformations of spacetime codes
No full textRecent advances in quantum error-correction (QEC) have shown that it is often beneficial to understand fault-tolerance as a dynamical process, a circuit with redundant measurements that help correct errors, rather than as a static code equipped with a syndrome extraction circuit. Spacetime codes have emerged as a natural framework to understand error correction at the circuit level while leveraging the traditional QEC toolbox. Here, we introduce a framework based on chain complexes and chain maps to model spacetime codes and transformations between them. We show that stabilizer codes, quantum circuits, and decoding problems can all be described using chain complexes, and that the equivalence of two spacetime codes can be characterized by specific maps between chain complexes, the fault-tolerant maps, that preserve the number of encoded qubits, fault distance, and minimum-weight decoding problem. As an application of this framework, we extend the foliated cluster state construction from stabilizer codes to any spacetime code, showing that any Clifford circuit can be transformed into a measurement-based protocol with the same fault-tolerant properties. To this protocol, we associate a chain complex which encodes the underlying decoding problem, generalizing previous cluster state complex constructions. Our method enables the construction of cluster states from non-CSS, subsystem, and Floquet codes, as well as from logical Clifford operations on a given code
CLAImate: AI-Enabled Climate Change Communication through Personalized and Localized Narrative Visualizations
No full textInternational audienceCommunicating climate change remains challenging, as climate reports, though rich in data and visualizations, often feel too abstract or technical for the public. Although personalization can enhance communication, most tools still lack the narrative and visualization tailoring needed to connect with individual experiences. We present CLAImate, an AI-enabled prototype that personalizes conversation narratives and localizes visualizations based on users’ climate knowledge and geographic location. We evaluated CLAImatethrough internal verification of factual correctness, a formative study with experts, and a pilot with UK residents. CLAImate achieved 66% SNLI accuracy and 70% FACTSCORE. Visualization experts appreciated its clarity and personalization, and seven out of ten UK participants reported better understanding and local relevance of climate risks with CLAImate. We also discuss design challenges in personalization, accuracy, and scalability, and outline future directions for integrating visualizations in personalized conversational interfaces
Proof Verification with GDV and LambdaPi - It's a Matter of Trust
No full textInternational audienceAutomated Theorem Proving (ATP) is concerned with the development and use of software that automates sound reasoning. An ATP system can be required to output a proof that serves as a certificate for the system's claim. To ensure that a proof is correct, verification can be required. If the verifier outputs evidence in a form that can be independently checked, that evidence serves as a certificate for the verifier's claim. The sequence of finding a proof, verifying the proof, and certifying the verification, builds an increasing level of trust in the system. This paper traces one such path for TPTP format proofs generated by ATP systems, via the GDV derivation verifier, and ending at the LambdaPi checker
Distributed On-Demand Routing for VLEO Constellations with 3-Terminal Inter-Satellite Links
No full textInternational audienceThis paper investigates routing strategies for Very Low Earth Orbit (VLEO) satellite constellations constrained by a three-terminal (3T) inter-satellite architecture, which is gaining interest as currently adopted by major operators (e.g., Starlink). In the 3T configuration, two terminals enable alongtrack communication, while a third hemispherical terminal can dynamically connect to either the left or right cross-plane neighbor, but not both simultaneously. This contrasts with the traditional four-terminal (4T) architecture, where four directional terminals enable full inter-plane links. The 3T constraint reduces crossplane connectivity, impacting path diversity, reachability, and latency. To address these challenges, we develop Interleaved-Scenario MinHopCount (IS-MHC) and DisCo3T, modified versions of the state-of-the-art routing algorithms Minimum Hop Count (MinHopCount) and Distributed On-Demand Routing for Mega-Constellations (DisCoRoute) for the 3T case. Our new model accurately captures the limitations of 3T connectivity and reduces hop count in this scenario. We conduct a comprehensive evaluation of DisCo3T in terms of latency, hop count, and computational complexity, including a comparative analysis with the original 4T implementation. Results show that the topological constraints of the 3T configuration cause an average hop count increase of 21.7% compared to the 4T baseline; however, DisCo3T retains the processing efficiency of the original 4T algorithm