Portail HAL des publications du LIRMM
Not a member yet
13279 research outputs found
Sort by
Hypergraph Clustering with Path-Length Awareness
No full textInternational audienceElectronic design automation toolchains require solving various circuit manipulation problems, such as floor planning, placement and routing. These circuits may be implemented using either Very Large-Scale Integration (VLSI) or Field Programmable Gate Arrays (FPGAs). However, with the ever-increasing size of circuits, now up to billions of gates, straightforward approaches to these problems do not scale well. A possible approach to reduce circuit complexity is to cluster circuits. In this work, we consider the problem of clustering combinatorial circuits, without cell replication. We propose a dedicated clustering algorithm based on binary search and study and improve the existing parameterized approximation ratio from M² + M (with M being the maximum size of each cluster) to M under specific hypothesis. We present an extension of the weighting schemes to model path length more accurately. This weighting scheme is combined with clustering methods based on a recursive matching algorithm. We evaluate and compare our approximation algorithm and recursive matching on several circuit instances and we obtain better results for a large number of instances with our algorithm than recursive matching
Towards package opening detection at power-up by monitoring thermal dissipation
No full textAmong the various threats to secure ICs, many are semi-invasive in the sense that their application requires the removal of the package to gain access to either the front or back of the target IC. Despite this stringent application requirements, little attention is paid to embedded techniques aiming at checking the package's integrity. This paper explores the feasibility of verifying the package integrity of microcontrollers by examining their thermal dissipation capability
Assegurando a Confidencialidade de Dados de Workflows Executados em Nuvens de Computadores: Abordagens Heurísticas e Exatas
No full textInternational audienceCloud computing provides an on-demand environment that allows users to execute their local workflows in an elastic and highly available environment. Various applications can be mode- led as workflows, many of which are compute- and data-intensive. In the cloud, the storage location of this data becomes a concern when confidentiality might be compromised. Malicious users can make inferences about the results and the structure of the workflows. Data dispersion, encryption, and other mechanisms can be adopted to enhance data privacy, but these cannot be implemented without considering the workflow scheduling, as this risks significantly increasing execution time and financial cost. In this paper, we introduce CYCLOPS, an approach that aims to execute work- flows in cloud computing environments efficiently while considering the confidentiality constraints of the produced data and the workflow structure.As nuvens de computadores fornecem um ambiente sob demanda que permite aos usuários executar seus workflows locais em um ambiente elástico e com alta disponibilidade. Diversas aplicac ¸ões podem ser modeladas como workflows, e muitas delas são intensivas em computac ¸ão e produc ¸ão de dados. Na nuvem, o local de armazenamento desses dados se torna uma preocupac ¸ão quando a confidencialidade pode ser comprometida. Usuários maliciosos podem realizar inferências a respeito dos resultados e da própria estrutura dos workflows. A dispersão de dados, a criptografia e outros mecanismos podem ser adotados para aprimorar a privacidade dos dados, mas estes não podem ser adotados sem considerar o escalonamento do workflow, pois isso arrisca aumentar significativamente o tempo de execuc ¸ão e o custo financeiro. Neste artigo, introduzimos a CYCLOPS, uma abordagem que visa executar workflows em nuvens de computadores de forma eficiente levando em considerac ¸ão as restric ¸ões de confidencialidade dos dados produzidos e da estrutura do workflow
M4.4 - Review of Semantic Artefact Catalogues and guidelines for serving FAIR semantic artefacts in EOSC
No full textIn the rapidly evolving landscape of scientific research, the proliferation of ontologies and semantic artefacts necessitates the development of robust systems to manage and utilise these resources effectively. Semantic Artefact Catalogues (SAC) and ontology repositories are critical in this regard, especially within the framework of the European Open Science Cloud (EOSC) that has clearly identified the important role “ontologies and metadata” may have on the construction of a Web of FAIR data and services. These catalogues provide essential platforms for receiving, hosting, serving, aligning, and enabling the reuse of ontologies and other Semantic Artefacts (SA) (terminologies, taxonomies, thesauri, vocabularies, metadata schemas and standards). These catalogues not only facilitate the organisation and access of semantic artefacts but also support and sometime ensure their compliance with the FAIR (Findable, Accessible, Interoperable, and Reusable) Data Principles, which are foundational to the EOSC’s mission of promoting open science and data sharing across diverse scientific disciplines. Semantic artefact catalogues essentially help their users to discover, manipulate, explore and exploit SAs without the need to manage or develop them. The types of SACs build by various (scientific but not only) communities ranges from simple semantic artefact listings to rich libraries with structured metadata, and advanced repositories (or portals) that offer a variety of services for multiple types of semantic artefacts,. These services may include browsing/searching, visualisation, metrics, recommendations, and annotation of data. SAC are often developed or maintained by specific discipline communities or infrastructures and we have seen the emergence of specific generic technologies –such as OntoPortal, SKOSMOS or OLS– that can be reused to deploy new semantic artefact catalogues. Within FAIR-IMPACT’s WP4 on ontologies and metadata, T4.2 aims to establish guidelines and community practices with respect to the lifecycle of FAIR semantic artefacts from creation (T4.2.1) to sharing and reuse via catalogues or repositories (T4.2.2) and standardisation of SA metadata descriptions and SAC application programming interfaces. WP4 has already produced multiple deliverables showing the importance of SACs in the governance of semantic artefacts (M4.1, D4.1) and in their FAIR lifecycle (M4.2). In this Milestone, we explore the current landscape of SACs, in EOSC and beyond; we make a quite comprehensive review of current and past SACs, sorting them by types, disciplines and technology. Plus, based on the five methodologies and tools for FAIRness assessment of SAs, available thru FAIR-IMPACT’s partners (O’FAIRe, FOOPS!, FsF, 10-SR and FVF) we have regrouped 10 important dimensions for FAIR semantic artefacts and we study how much each reviewed SAC enables or supports FAIR for their artefacts. The Milestone provides a good overview of how SACs can help SAs to address FAIR principles and contribute to the efficient management and utilisation of SAs. The Milestone consists of the current report presenting our methodology and result analysis as well as associated data under the form of a spreadsheet which contains the listing of SACs, their classifications (status, type, discipline, technology) and the evaluation of their FAIR-enabling dimensions. The spreadsheet discussed and analysed in the current report is versioned with DOI: 10.5281/zenodo.1279986
Collaborative Benchmarking Rule-Reasoners with B-Runner
No full textInternational audienceConducting experiment alanalysis on rule reasoners is a mainstream task for validating novel algorithms and systems. Nevertheless, providing robust, verifiable, and reproducible experiments can still raise a sensible challenge. This paper introduces B-Runner, an open library for collaborative benchmarking focusing on the deployment of extended tests for knowledge and rule-based systems with low cost and high robustness. B-Runner reduces the benchmarking setup time while guaranteeing ex- periment repeatability. Also, it improves the scrutability of experimental protocols thereby enhancing the fairness of system comparisons
Mass Spectrometry-Based Pipeline for Identifying RNA Modifications Involved in a Functional Process: Application to Cancer Cell Adaptation
No full textInternational audienceCancer onset and progression are known to be regulated by genetic and epigenetic events, including RNA modifications (a.k.a. epitranscriptomics). So far, more than 150 chemical modifications have been described in all RNA subtypes, including messenger, ribosomal, and transfer RNAs. RNA modifications and their regulators are known to be implicated in all steps of post-transcriptional regulation. The dysregulation of this complex yet delicate balance can contribute to disease evolution, particularly in the context of carcinogenesis, where cells are subjected to various stresses. We sought to discover RNA modifications involved in cancer cell adaptation to inhospitable environments, a peculiar feature of cancer stem cells (CSCs). We were particularly interested in the RNA marks that help the adaptation of cancer cells to suspension culture, which is often used as a surrogate to evaluate the tumorigenic potential. For this purpose, we designed an experimental pipeline consisting of four steps: (1) cell culture in different growth conditions to favor CSC survival; (2) simultaneous RNA subtype (mRNA, rRNA, tRNA) enrichment and RNA hydrolysis; (3) the multiplex analysis of nucleosides by LC-MS/MS followed by statistical/bioinformatic analysis; and (4) the functional validation of identified RNA marks. This study demonstrates that the RNA modification landscape evolves along with the cancer cell phenotype under growth constraints. Remarkably, we discovered a short epitranscriptomic signature, conserved across colorectal cancer cell lines and associated with enrichment in CSCs. Functional tests confirmed the importance of selected marks in the process of adaptation to suspension culture, confirming the validity of our approach and opening up interesting prospects in the field
Improved Adaptive High-Order Sliding Mode-Based Control for Trajectory Tracking of Autonomous Underwater Vehicles
No full textInternational audienceWhen an autonomous underwater vehicle is performing missions in the ocean, it is often subject to external disturbances, such as sea currents and changes in salinity. These phenomena can degrade the performance of the controller of the vehicle, which can increase the tracking error or cause instability. Taking into account these issues, in this article, we design an adaptive controller based on high-order sliding mode control and focus on the paradigm of the trajectory tracking of underwater vehicles. First, we rewrite the classical representation of the underwater vehicle in terms of the known dynamics and then transform it into a control affine structure. Then, we design an adaptive high-order sliding mode controller for trajectory tracking. Also, we prove the stability of the resulting closed-loop system using Lyapunov arguments. Finally, real-time experiments are performed to validate the proposed methodology, as well as its robustness
Analyzing GPU Energy Consumption in Data Movement and Storage
No full textInternational audienceGPUs are the prevailing solution to execute high- performance tasks (e.g., machine learning training). As the peak performance of modern GPUs increases with each generation, so does their thermal design power (TDP). Hence, identifying energy bottlenecks in the GPU architecture is crucial to designing more efficient architectures in the future. However, due to the complex proprietary nature of modern GPU architectures, providing a detailed breakdown of the GPU energy consumption is not trivial.The goal of this work is to estimate a lower bound for the energy consumed by data movement and storage in modern GPU architectures, leveraging internal power sensors. We establish a basic energy model for modern GPUs, focused on data movement to/from the hardware-managed caches and software-managed memories. We propose a methodology to calibrate the energy model using microbenchmarks, performance counters, and the internal power sensor. We experimentally calibrate the model on an A100 NVIDIA GPU. Then, we challenge the consistency of the results by cross-validating with modified microbenchmarks with additional instructions. Finally, we use the calibrated energy model to evaluate breakdowns for workloads of increasing complexity (e.g., a ResNet-50 training iteration with different software optimizations). Our results show that data movement dominates the dynamic energy consumption of the GPU (up to 84%), with DRAM accesses being the main contributor
Réseaux de Feistel Généralisés pour Masquage Efficace dans les Corps Premiers
No full textInternational audienceA recent work from Eurocrypt 2023 suggests that prime-field masking has excellent potential to improve the efficiency vs. security tradeoff of masked implementations against side-channel attacks, especially in contexts where physical leakages show low noise. We pick up on the main open challenge that this seed result leads to, namely the design of an optimized prime cipher able to take advantage of this potential. Given the interest of tweakable block ciphers with cheap inverses in many leakage-resistant designs, we start by describing the FPM (Feistel for Prime Masking) family of tweakable block ciphers based on a generalized Feistel structure. We then propose a first instantiation of FPM, which we denote as small-pSquare. It builds on the recent observation that the square operation (which is non-linear in Fp) can lead to masked gadgets that are more efficient than those for multiplication, and is tailored for efficient masked implementations in hardware. We analyze the mathematical security of the FPM family of ciphers and the small-pSquare instance, trying to isolate the parts of our study that can be re-used for other instances. We additionally evaluate the implementation features of small-pSquare by comparing the efficiency vs. security tradeoff of masked FPGA circuits against those of a state-of-the art binary cipher, namely SKINNY, confirming significant gains in relevant contexts
Checkpointing optimal pour chaînes hétérogènes: apprentissage de réseaux de neurones profonds avec mémoire limitée
No full textInternational audienceTraining in Feed Forward Deep Neural Networks is a memory-intensive operation which is usually performed on GPUs with limited memory capacities. This may force data scientists to limit the depth of the models or the resolution of the input data if data does not fit in the GPU memory. The re-materialization technique, whose idea comes from the checkpointing strategies developed in the Automatic Differentiation literature, allows data scientists to limit the memory requirements related to the storage of intermediate data (activations), at the cost of an increase in the computational cost.This paper introduces a new strategy of re-materialization of activations that significantly reduces memory usage. It consists in selecting which activations are saved and which activations are deleted during the forward phase and then recomputing the deleted activations when they are needed during the backward phase.We propose an original computation model that combines two types of activation savings: either only storing the layer inputs or recording the complete history of operations that produced the outputs. This paper focuses on the fully heterogeneous case, where the computation time and the memory requirement of each layer is different. We prove that finding the optimal solution is NP-hard and that classical techniques from Automatic Differentiation literature do not apply. Moreover, the classical assumption of memory persistence of materialized activations, used to simplify the search of optimal solutions, does not hold anymore. Thus, we propose a weak memory persistence property and provide a dynamic program to compute the optimal sequence of computations.This algorithm is made available through the Rotor software, a PyTorch plug-in dealing with any network consisting of a sequence of layers, each of them having an arbitrarily complex structure. Through extensive experiments, we show that our implementation consistently outperforms existing re-materialization approaches for a large class of networks, image sizes, and batch sizes