arXiv subjects

David Mercier

Publications and source records attributed to David Mercier.

7 recordsLinked to original sources

arXiv2026

Improving Multivariate Time Series Classification with Class-Wise Training and Model Aggregation

In this paper, we propose a class-wise dimension (channel) selection framework for Multivariate Time Series Classification (MTSC). Rather than applying a single global dimension selection process, the proposed approach independently identifies informative dimensions for each class. A dedicated learning process is subsequently performed for each class, followed by a fusion stage for final prediction. The objective is to improve the generation of discriminative feature representations while reducing the influence of noisy or non-informative dimensions. The proposed framework is evaluated using MiniRocket, a random kernel-based baseline method. Experimental results indicate that class-wise dimension selection improves the quality of extracted representations and can enhance classification performance, particularly in high-dimensional settings. These findings suggest that incorporating class-specific information into the training process represents a promising direction for MTSC, improving robustness through consistent gains across heterogeneous datasets, and interpretability through the explicit identification of class-relevant dimensions.

cs.LG↗

arXiv2026

Multi-Class Electrical and Mechanical Fault Classification Using Random Convolutional Kernels

Diagnosing faults in rotating machinery is essential for ensuring the reliability of industrial processes. Random convolutional kernel-based Time Series Classification (TSC) methods, such as ROCKET and its variants, provide an attractive trade-off between predictive performance and computational efficiency. In this work, we evaluate SelF-Rocket for the multi-class diagnosis of both mechanical and electrical faults and introduce, as a new contribution, a multivariate extension of the original method. The proposed approach is compared with leading ROCKET-based methods on two public benchmark datasets, MaFaulDa (mechanical faults) and ITSC-UDG (stator inter-turn short circuits), under both univariate and multivariate settings. Experimental results show that SelF-Rocket achieves the best overall accuracy-latency trade-off among the evaluated methods, obtaining the highest classification performance on MaFaulDa while remaining highly competitive on the more challenging ITSC-UDG dataset.

cs.LG↗

arXiv2026

Enhancing Phase Clustering in Nanomechanical Property Maps of Multiphase Materials Using Kernel-Averaged Mechanical Mismatch

This work presents a novel approach to improve phase identification in nanomechanical property maps of multiphase materials, such as those obtained by nanoindentation or atomic force microscopy (AFM). A major challenge in validating clustering strategies for these data is the lack of ground-truth phase labels in experimental measurements, along with the tendency of overly simplistic synthetic datasets to artificially inflate algorithm performance. To address this gap, we construct controlled yet non-trivial synthetic benchmarks with tunable mechanical contrast, graded interfaces, curved boundaries, and diffuse morphologies, enabling rigorous and realistic evaluation of clustering robustness. Conventional clustering methods based only on elastic modulus (E) and hardness (H) often struggle to separate phases when mechanical contrast is low or when diffuse interphase regions are present. We introduce the Kernel-Averaged Mechanical Mismatch (KAMM), a neighborhood-informed feature that quantifies local mechanical heterogeneity by comparing each point to its neighbors in (E, H) space. When incorporated into a three-dimensional clustering space (E, H, KAMM), this framework improves phase separability, enhances interphase detection, and increases robustness to noise. By enabling more reliable segmentation of mechanical domains under realistic contrast conditions, the proposed method facilitates the generation of representative volume elements (RVEs) and supports more accurate extraction of phase-specific properties in heterogeneous microstructures.

cond-mat.mtrl-sci↗

arXiv2026

A Curated Literature Database for Monitoring More Than 30 Years of Ansys Granta Product Usage

Engineering and materials software is increasingly difficult to track in the scholarly and technical literature because publication volume is growing rapidly and software citation practices remain inconsistent. This is particularly true for the Ansys Granta product family, which is used for materials education, materials and process selection, sustainability-driven design, and enterprise materials information management. We present a structured and reproducible framework to consolidate evidence of \emph{operational} Granta usage and to support quantitative monitoring of adoption patterns, application domains, and technical impact. The framework is implemented as a curated reference database in \textit{Ansys Granta MI Enterprise}: bibliographic metadata are ingested semi-automatically (e.g., via DOI and citation-file parsing) and complemented by expert curation of usage descriptors (product, context, application domain, and technical depth), with relational links to authors and institutions. Downstream analytics are performed with Python, dashboards, and bibliometric/network visualization tools to enable reproducible querying and reporting. As of September~2025, the database contains more than 1{,}100 curated records spanning journals, conferences, theses, books, patents, standards, and reports, and supports rapid retrieval of validated case studies, reproducible literature reviews, and technology scouting. Example analyses highlight dominant domains, key institutions, and recurring integrations with CAD/CAE/FEM environments. Overall, the approach converts heterogeneous software-usage evidence into structured, analyzable knowledge to improve visibility of engineering software impact and to support evidence-based assessment and strategic decision-making.

cs.DL↗

arXiv2024

Time series classification with random convolution kernels: pooling operators and input representations matter

This article presents a new approach based on MiniRocket, called SelF-Rocket, for fast time series classification (TSC). Unlike existing approaches based on random convolution kernels, it dynamically selects the best couple of input representations and pooling operator during the training process. SelF-Rocket achieves state-of-the-art accuracy on the University of California Riverside (UCR) TSC benchmark datasets.

cs.LG↗

arXiv2019

SPSC: a new execution policy for exploring discrete-time stochastic simulations

In this paper, we introduce a new method called SPSC (Simulation, Partitioning, Selection, Cloning) to estimate efficiently the probability of possible solutions in stochastic simulations. This method can be applied to any type of simulation, however it is particularly suitable for multi-agent-based simulations (MABS). Therefore, its performance is evaluated on a well-known MABS and compared to the classical approach, i.e., Monte Carlo.

cs.MA↗

arXiv2018

Statistical and probabilistic modeling of a cloud of particles coupled with a turbulent fluid

This paper exposes a novel exploratory formalism, which end goal is the numerical simulation of the dynamics of a cloud of particles weakly or strongly coupled with a turbulent fluid. Giventhe large panel of expertise of the list of authors, the content of this paper scans a wide range of connexnotions, from the physics of turbulence to the rigorous definition of stochastic processes. Our approachis to develop reduced-order models for the dynamics of both carrying and carried phases which remainconsistant within this formalism, and to set up a numerical process to validate these models. Thenovelties of this paper lie in the gathering of a large panel of mathematical and physical definitionsand results within a common framework and an agreed vocabulary (sections 1 and 2), and in somepreliminary results and achievements within this context, section 3. While the first three sections havebeen simplified to the context of a gas field providing that the disperse phase only retrieves energythrough drag, the fourth section opens this study to the more complex situation when the dispersephase interacts with the continuous phase as well, in an energy conservative manner. This will allowus to expose the perspectives of the project and to conclude.

math.AP↗