检索结果-内蒙古大学图书馆

Multiprocessor Computing based Parallel Structures of Mathematical Models of Tridiagonal Systems

学校读者我要写书评

暂无评论

Multiprocessor Computing based Parallel Structures of Mathem...

International Conference on Inventive Computation Technologies (ICICT)

作者： Gennady Shvachych Ivan Pobochii Tetyana Khokhlova Olena Kholod Volodymyr Busygin Dmytro Moroz Department of Applied Mathematics and Computer Science National Metallurgical Academy of Ukraine Dnipro Ukraine Economics and Modeling Business Processes Alfred Nobel University Dnipro Ukraine Department of Economic Cybernetics Oles Honchar Dnipro National University Dnipro Ukraine

ISBN: (数字)9781728146850

ISBN: (纸本)9781728146867

The paper is devoted to the construction of parallel forms of mathematical models of a tridiagonal structure. Classical methods for solving mathematical models of a tridiagonal structure with multiprocessor computing systems are processed, as a rule, much slower than those with single-processor computing. This circumstance is explained by the recursive approach, which is the basis of classical methods. In this regard, this paper's primary purpose is to construct the most parallel forms when simulating the indicated class of problems. It is convenient to apply the parallelization algorithm and its mapping to parallel computing systems on the two schemes proposed in this paper: finite-difference and numerical-analytical. This approach allows arranging separate determination of the thermal physical characteristics of the structures' material, i.e., allows obtaining solutions of coefficient and other inverse problems of thermal *** paper considers two methods of discretization of differential problems on the example of solving the mathematical physics equation. Moreover, the application of the numerical-analytical straight-line method and sweep methods for parallelization of mathematical models with a three-diagonal structure allows constructing its exact node-by-node solutions having the most parallel form and the least implementation time on parallel computing ***, it is proposed to use finite-difference and numerical-analytical methods in combination with the splitting method as a methodological basis for constructing numerical methods for solving such problems. The splitting method provides an economical and sustainable implementation of numerical models by the scalar sweep *** proposed approach for the development of methods, algorithms, and software can be applied in various branches of metallurgical thermal physics, economics, as well as environmental problems of the metallurgical industry.

关键词：

Democratizing Uncertainty Quantification

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Seelinger, Linus Reinarz, Anne Lykkegaard, Mikkel B. Akers, Robert Alghamdi, Amal M.A. Aristoff, David Bangerth, Wolfgang Bénézech, Jean Diez, Matteo Frey, Kurt Jakeman, John D. Jørgensen, Jakob S. Kim, Ki-Tae Kent, Benjamin M. Martinelli, Massimiliano Parno, Matthew Pellegrini, Riccardo Petra, Noemi Riis, Nicolai A.B. Rosenfeld, Katherine Serani, Andrea Tamellini, Lorenzo Villa, Umberto Dodwell, Tim J. Scheichl, Robert Scientific Computing Center Karlsruhe Institute of Technology Karlsruhe Germany Department of Computer Science Durham University Durham United Kingdom digiLab Exeter United Kingdom UK Atomic Energy Authority Oxford United Kingdom Lyngby Denmark Department of Mathematics Colorado State University Fort CollinsCO United States Department of Mathematics Department of Geosciences Colorado State University Fort CollinsCO United States Centre for Integrated Materials Processes and Structures Department of Mechanical Engineering University of Bath Bath United Kingdom National Research Council Institute of Marine Engineering Rome Italy Institute for Disease Modeling Global Health Division Bill & Melinda Gates Foundation United States Optimization and Uncertainty Quantification Sandia National Laboratories AlbuquerqueNM United States University of California Merced United States National Research Council-Institute for Applied Mathematics and Information Technologies "E. Magenes" Pavia Italy Solea Energy ThetfordVT United States Copenhagen Imaging ApS Herlev Denmark The University of Texas Austin United States Heidelberg University Germany

Uncertainty Quantification (UQ) is vital to safety-critical model-based analyses, but the widespread adoption of sophisticated UQ methods is limited by technical complexity. In this paper, we introduce UM-Bridge (the UQ and modeling Bridge), a high-level abstraction and software protocol that facilitates universal interoperability of UQ software with simulation codes. It breaks down the technical complexity of advanced UQ applications and enables separation of concerns between experts. UM-Bridge democratizes UQ by allowing effective interdisciplinary collaboration, accelerating the development of advanced UQ methods, and making it easy to perform UQ analyses from prototype to High Performance Computing (HPC) scale. In addition, we present a library of ready-to-run UQ benchmark problems, all easily accessible through UM-Bridge. These benchmarks support UQ methodology research, enabling reproducible performance comparisons. We demonstrate UM-Bridge with several scientific applications, harnessing HPC resources even using UQ codes not designed with HPC support. © 2024, CC BY-NC-ND.

关键词： Application programs

Calibrated multi-probabilistic prediction as a defense against adversarial attacks 31st

学校读者我要写书评

暂无评论

Calibrated multi-probabilistic prediction as a defense again...

31st Benelux Conference on Artificial Intelligence and the 28th Belgian Dutch Conference on Machine Learning, BNAIC/BENELEARN 2019

作者： Peck, Jonathan Goossens, Bart Saeys, Yvan Department of Applied Mathematics Computer Science and Statistics Ghent University Ghent9000 Belgium Data Mining and Modeling for Biomedicine VIB Inflammation Research Center Ghent9052 Belgium Department of Telecommunications and Information Processing IMEC Ghent University Ghent9000 Belgium

ISBN: (纸本)9783030651534

Machine learning (ML) classifiers—in particular deep neural networks—are surprisingly vulnerable to so-called adversarial examples. These are small modifications of natural inputs which drastically alter the output of the model even though no relevant features appear to have been modified. One explanation that has been offered for this phenomenon is the calibration hypothesis, which states that the probabilistic predictions of typical ML models are miscalibrated. As a result, classifiers can often be very confident in completely erroneous predictions. Based on this idea, we propose the MultIVAP algorithm for defending arbitrary ML models against adversarial examples. Our method is inspired by the inductive Venn-ABERS predictor (IVAP) technique from the field of conformal prediction. The IVAP enjoys the theoretical guarantee that its predictions will be perfectly calibrated, thus addressing the problem of miscalibration. Experimental results on five image classification tasks demonstrate empirically that the MultIVAP has a reasonably small computational overhead and provides significantly higher adversarial robustness without sacrificing accuracy on clean data. This increase in robustness is observed both against defense-oblivious attacks as well as a defense-aware white-box attack specifically designed for the MultIVAP. We make our code available at https://***/saeyslab/multivap. © 2020, Springer Nature Switzerland AG.

关键词： Forecasting

sbi reloaded: a toolkit for simulation-based inference workflows

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Boelts, Jan Deistler, Michael Gloeckler, Manuel Tejero-Cantero, Álvaro Lueckmann, Jan-Matthis Moss, Guy Steinbach, Peter Moreau, Thomas Muratore, Fabio Linhart, Julia Durkan, Conor Vetter, Julius Miller, Benjamin Kurt Herold, Maternus Ziaeemehr, Abolfazl Pals, Matthijs Gruner, Theo Bischoff, Sebastian Krouglova, Anastasia N. Gao, Richard Lappalainen, Janne K. Mucsányi, Bálint Pei, Felix Schulz, Auguste Stefanidi, Zinovia Rodrigues, Pedro L.C. Schröder, Cornelius Zaid, Faried Abu Beck, Jonas Kapoor, Jaivardhan Greenberg, David S. Gonçalves, Pedro J. Macke, Jakob H. Machine Learning in Science University of Tübingen Germany Tübingen AI Center Germany TransferLab AppliedAI Institute for Europe Germany ML Colab Cluster ML in Science University of Tübingen Germany Google Research United States Helmholtz-Zentrum Dresden-Rossendorf Germany Université Paris-Saclay INRIA CEA Palaiseau France Robert Bosch GmbH Germany School of Informatics University of Edinburgh United Kingdom University of Amsterdam Netherlands Research and Innovation Center BMW Group Germany Institute for Applied Mathematics and Scientific Computing University of the Bundeswehr Munich Germany Aix Marseille INSERM INS France TU Darmstadt Hessian.AI Germany University Hospital Tübingen M3 Research Center Germany Faculty of Science KU Leuven B-3000 Belgium Imec Belgium Methods of Machine Learning University of Tübingen Germany Neuroscience Institute Carnegie Mellon University United States Université Grenoble Alpes INRIA CNRS Grenoble INP LJK France Hertie Institute for AI in Brain Health University of Tübingen Germany Institute of Coastal Systems - Analysis and Modeling Helmholtz AI Germany Departments of Computer Science Electrical Engineering KU Leuven Belgium Department Empirical Inference Max Planck Institute for Intelligent Systems Tübingen Germany

Scientists and engineers use simulators to model empirically observed phenomena. However, tuning the parameters of a simulator to ensure its outputs match observed data presents a significant challenge. Simulation-based inference (SBI) addresses this by enabling Bayesian inference for simulators, identifying parameters that match observed data and align with prior knowledge. Unlike traditional Bayesian inference, SBI only needs access to simulations from the model and does not require evaluations of the likelihood-function. In addition, SBI algorithms do not require gradients through the simulator, allow for massive parallelization of simulations, and can perform inference for different observations without further simulations or training, thereby amortizing inference. Over the past years, we have developed, maintained, and extended sbi, a PyTorch-based package4 that implements Bayesian SBI algorithms based on neural networks. The sbi toolkit implements a wide range of inference methods, neural network architectures, sampling methods, and diagnostic tools. In addition, it provides well-tested default settings but also offers flexibility to fully customize every step of the simulation-based inference workflow. Taken together, the sbi toolkit enables scientists and engineers to apply state-of-the-art SBI methods to black-box simulators, opening up new possibilities for aligning simulations with empirically observed data. © 2024, CC BY-SA.

关键词： Neural networks

Inline detection of DGA domains using side information

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Sivaguru, Raaghavi Peck, Jonathan Olumofin, Femi Nascimento, Anderson de Cock, Martine School of Engineering and Technology University of Washington Tacoma United States Department of Applied Mathematics Computer Science and Statistics Ghent University Ghent9000 Belgium Data Mining and Modeling for Biomedicine VIB Inflammation Research Center Ghent9052 Belgium Infoblox Santa Clara United States

Malware applications typically use a command and control (C&C) server to manage bots to perform malicious activities. Domain Generation Algorithms (DGAs) are popular methods for generating pseudo-random domain names that can be used to establish a communication between an infected bot and the C&C server. In recent years, machine learning based systems have been widely used to detect DGAs. There are several well known state-of-the-art classifiers in the literature that can detect DGA domain names in real-time applications with high predictive performance. However, these DGA classifiers are highly vulnerable to adversarial attacks in which adversaries purposely craft domain names to evade DGA detection classifiers. In our work, we focus on hardening DGA classifiers against adversarial attacks. To this end, we train and evaluate state-of-the-art deep learning and random forest (RF) classifiers for DGA detection using side information that is harder for adversaries to manipulate than the domain name itself. Additionally, the side information features are selected such that they are easily obtainable in practice to perform inline DGA detection. The performance and robustness of these models is assessed by exposing them to one day of real-traffic data as well as domains generated by adversarial attack algorithms. We found that the DGA classifiers that rely on both the domain name and side information have high performance and are more robust against adversaries. Copyright © 2020, The Authors. All rights reserved.

关键词： Malware

Formation of Space Charge Wave with Wide Frequency Spectrum in Helical Electron Beam of Two-Stream FEL Transit Section with Longitudinal Electric Field

学校读者我要写书评

暂无评论

Formation of Space Charge Wave with Wide Frequency Spectrum ...

International Conference on Advanced Optoelectronics and Lasers (CAOL)

作者： Alexander Lysenko Iurii Volk Galyna Oleksiienko Department of Applied Mathematics and Complex Systems Modeling Sumy State University Sumy Ukraine Department of Computer Science Sumy State University Sumy Ukraine

ISBN: (数字)9781728118147

ISBN: (纸本)9781728118154

The longitudinal electric field effect on the formation of a space charge wave (SCW) with wide frequency spectrum in the helical two-stream relativistic electron beam (REB), which moves in transit section of multiharmonic two-stream superheterodyne free-electron laser, is analyzed. In the framework of cubic nonlinear approximation, it is shown that the use of the accelerating electric field makes it possible to increase the multiharmonic SCW width of the frequency spectrum. It is shown that the retarding electric field causes the increase of the amplitude level SCW harmonics, as well as the decrease of its saturation length.

关键词： Harmonic analysis Electric fields Electron beams Resonant frequency Mathematical model Acceleration Space charge

Regional Image Perturbation Reduces Lp Norms of Adversarial Examples While Maintaining Model-to-model Transferability

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Ozbulak, Utku Peck, Jonathan de Neve, Wesley Goossens, Bart Saeys, Yvan van Messem, Arnout Department of Electronics and Information Systems Ghent University Ghent Belgium Center for Biotech Data Science Ghent University Global Campus Incheon Korea Republic of Department of Applied Mathematics Computer Science and Statistics Ghent University Ghent Belgium Data Mining and Modeling for Biomedicine VIB Inflammation Research Center Ghent Belgium Department of Telecommunications and Information Processing Ghent University - imec Ghent Belgium

Regional adversarial attacks often rely on complicated methods for generating adversarial perturbations, making it hard to compare their efficacy against well-known attacks. In this study, we show that effective regional perturbations can be generated without resorting to complex methods. We develop a very simple regional adversarial perturbation attack method using cross-entropy sign, one of the most commonly used losses in adversarial machine learning. Our experiments on ImageNet with multiple models reveal that, on average, 76% of the generated adversarial examples maintain model-to-model transferability when the perturbation is applied to local image regions. Depending on the selected region, these localized adversarial examples require significantly less Lp norm distortion (for p ∈ {0, 2, ∞}) compared to their non-local counterparts. These localized attacks therefore have the potential to undermine defenses that claim robustness under the aforementioned norms. Copyright © 2020, The Authors. All rights reserved.

关键词： Perturbation techniques

Game theory of pandemic control: Can collaborative vaccine allocation strategies lead to better outcomes?

学校读者我要写书评

暂无评论

IJID Regions 2025年 14卷 100535-100535页

作者： Yiwen Wang Yang Ye Xiaoyi Su Qingpeng Zhang Hsiang-Yu Yuan Department of Computer Science and Engineering The Chinese University of Hong Kong Hong Kong China Center for Infectious Disease Modeling and Analysis Yale School of Public Health New Haven CT 06520 Department of Biostatistics City University of Hong Kong Hong Kong China HKU Musketeers Foundation Institute of Data Science and Department of Pharmacology and Pharmacy The University of Hong Kong Hong Kong China Department of Biomedical Sciences Jockey Club College of Veterinary Medicine and Life Sciences City University of Hong Kong Hong Kong China Centre for Applied One Health Research and Policy Advice Jockey Club College of Veterinary Medicine and Life Sciences City University of Hong Kong Hong Kong China

Introduction Increasing vaccine uptake is an effective way to prevent COVID-19 infection in individual countries after the relaxation of social distancing. However, under the limited vaccine resources, vaccine inequality between high- and low-income countries can also produce more infection and cost. Reducing such inequality relies on a more cooperative (unselfish) strategy, such as vaccine donation from high-income countries, which may initially increase their cost. The study aims to analyze the impact of vaccine inequality on cooperative decision-making between two populations. The results provide important insights on achieving global control of infectious disease outbreaks. Material and Methods We developed a transmission model incorporating virus evolution under different vaccine-induced natural selection and the migration to allow a long-term repeated outbreak. Vaccine allocation strategies were implemented between two populations the high vaccine coverage population (HC) and the low coverage population (LC). A game theory approach was employed to explore whether pursuing minimum cost for both sides (cooperative strategy) or itself (non-cooperative strategy) for each population. We estimated the effect of each strategy on the total cost (i.e. the sum of vaccine, hospitalization and labor costs). The Nash equilibrium was obtained by minimizing their own cost for individual countries under different vaccine inequality scenarios (from low to high inequality). Results Under low inequality, both populations achieved the same optimal cost with both cooperative and non-cooperative strategies. Under medium or high inequality, both populations reached a Nash equilibrium with non-cooperative strategies and could not reach global optimum. When inequality was high, HC tended towards vaccine donation even though non-cooperative strategies were adopted. Discussion When inequality is high, although cooperative decisions can obtain global optimal outcomes, populations tend to

关键词：

学校读者我要写书评

暂无评论

Preface

CEUR Workshop Proceedings 2022年 3091卷

作者： Kovalev, Igor Semenkin, Eugene Schwenker, Friedhelm Shaidurov, Vladimir Ganchev, Todor Škraba, Andrej Stanimirović, Predrag Kolehmainen, Mikko Kronberger, Gabriel Affenzeller, Michael Brester, Christina Ryzhikov, Ivan Kazakovtsev, Lev Sopov, Evgenii Semenkina, Maria Akhmedova, Shakhnaz Stanovov, Vladimir Krasnoyarsk Regional Science and Technology City Hall Department of Computer Science of Siberian Federal University Krasnoyarsk Russia Siberian Institute of Applied System Analysis named after A.N. Antamoshkin The Department ofsystem analysis and operations research of Reshetnev University Krasnoyarsk Russia Institute of Neural Information Processing Ulm University Ulm Germany The Institute of Computational Modeling SB RAS Krasnoyarsk Russia Technical University of Varna Varna Bulgaria University of Maribor UM Faculty of Organizational Sciences Cybernetics and Decision Support Systems Laboratory Slovenia Department of Computer Science Niš Serbia Department of Environmental and Biological Sciences Environmental Informatics University of Eastern Finland Kuopio Finland University of Applied Sciences Upper Austria Hagenberg Austria Department of Software Engineering School of Informatics Communications and Media University of Applied Sciences Upper Austria Hagenberg Austria University of Eastern Finland UEF · Department of Environmental Science Kuopio Finland The Department of Systems Analysis Reshetnev University of Science and Technology Krasnoyarsk Russia Siberian Institute of Applied System Analysis Krasnoyarsk Russia University of Applied Sciences Upper Austria and Softwarepark Hagenberg Austria Reshetnev Siberian States University of Science and Technology Siberian Institute of Applied System Analysis Krasnoyarsk Russia SibSAU · Institute of computer science and telecommunications Krasnoyarsk Russia