检索结果-内蒙古大学图书馆

Load Balancing Optimization for Transformer in Distributed Environment 27

学校读者我要写书评

暂无评论

Load Balancing Optimization for Transformer in Distributed E...

27th IEEE International Conference on Parallel and Distributed Systems, ICPADS 2021

作者： Ma, Delu Lei, Zhou Chen, Shengbo Wang, Peng School of Computer Engineering and Science Shanghai University Shanghai200444 China Shanghai Key Laboratory of Computer Software Testing and Evaluating Shanghai201112 China

ISBN: (纸本)9781665408783

In recent years, the demand for artificial intelligence applications has increased dramatically. Complex models can promote machine learning to achieve excellent results, but computing efficiency has gradually reached a bottleneck. Therefore, more researchers are exploring the improvement of the efficiency of intelligent computing systems. Distributed machine learning can improve the efficiency of model training and inference, but problems such as communication delay and load imbalance between computing nodes still exist. In the multi-GPU distributed computing environment, this paper takes the vision field algorithm VIT (vision transformer) as the optimization object, which has the advantage of convenient parallel training, and proposes several related solutions. Firstly, the parameter server is used as the system logic architecture and in order to reduce the idleness of the computing devices during the training process, the device working status query mechanism is designed to realize load balancing. Secondly, combined with the pre-trained small VIT algorithm model, semi-asynchronous communication method is proposed to reduce the communication overhead of computing devices and accelerate global convergence. The results of this experiment carried out in the existing distributed environment has demonstrated that compared with the existing synchronization method, the computational efficiency has been improved well under the premise of slightly reducing the accuracy. © 2021 IEEE.

关键词： computation theory

Analysis of Kuramoto-Sivashinsky Model of Flame/Smoldering Front by Means of Curvature Driven Flow

学校读者我要写书评

暂无评论

Analysis of Kuramoto-Sivashinsky Model of Flame/Smoldering F...

European Conference on Numerical Mathematics and Advanced Applications, ENUMATH 2019

作者： Kolář, Miroslav Kobayashi, Shunsuke Uegata, Yasuhide Yazaki, Shigetoshi Beneš, Michal Graduate School of Science and Technology Meiji University Kanagawa Japan Graduate School of Science Kyoto University Kyoto Japan Setagaya Gakuen School Tokyo Japan Department of Mathematics Faculty of Nuclear Sciences and Physical Engineering Czech Technical University in Prague Prague Czech Republic

ISBN: (纸本)9783030558734

In this paper we summarize our results on the investigation of the Kuramoto-Sivashinsky model, which describes the motion of flame/smoldering interface. We propose the generalization of the model formulated in terms of mathematical theory of moving parametrized curves, and investigate it from numerical and analytical point of view. In the part dedicated to computational studies, we present the verification of our scheme by measurement of experimental order of convergence. In the analytical part of the paper we summarize biffurcation analysis of the model and study of rotational wave solutions. © 2021, Springer Nature Switzerland AG.

关键词： computation theory

Optimized Transfer Learning: Application for Wireless Channel Selection 17

学校读者我要写书评

暂无评论

Optimized Transfer Learning: Application for Wireless Channe...

17th International Conference on Network and Service Management, CNSM 2021

作者： Askarizadeh, Mohammad Hussien, Mostafa Zare, Masoumeh Nguyen, Kim Khoa Université du Québec Dept. Electrical Engineering Éts MontréalQC Canada

ISBN: (纸本)9783903176362

Recently, transfer learning (TL) has emerged as a powerful machine learning method in distributed environments. Transferring the knowledge between distributed agents helps reduce both learning time and computing costs. However, in a communication system, the advantage of TL comes with communication costs. To make an optimal decision of transfer between two agents, we try to answer three key questions: i) which information should be transferred from a source to a target?, ii) how this transferred information will be adapted to the target? and iii) when should TL be triggered to optimize the costs?. To this end, we introduce a new concept of similarity based on the Best Approximation theory and a general transfer rule. Then, we propose a model to evaluate the feasibility and optimality of TL. We verify our proposed model in the context of the wireless channel selection problem using contextual multi-armed bandits. Experimental results show optimal TL decisions can be made, and Extra Action is an efficient technique for TL in channel selection. © 2021 IFIP.

关键词： computation theory

Confidential Consortium Framework: Secure Multiparty Applications with Confidentiality, Integrity, and High Availability

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Howard, Heidi Alder, Fritz Ashton, Edward Chamayou, Amaury Clebsch, Sylvan Costa, Manuel Delignat-Lavaud, Antoine Fournet, Cédric Jeffery, Andrew Kerner, Matthew Kounelis, Fotios Kuppe, Markus A. Maffre, Julien Russinovich, Mark Wintersteiger, Christoph M. Azure Research Microsoft United States imec-DistriNet KU Leuven Belgium Belgium University of Cambridge United Kingdom Microsoft United States Imperial College London United Kingdom Microsoft Research United States

Confidentiality, integrity protection, and high availability, abbreviated to CIA, are essential properties for trustworthy data systems. The rise of cloud computing and the growing demand for multiparty applications however means that building modern CIA systems is more challenging than ever. In response, we present the Confidential Consortium Framework (CCF), a general-purpose foundation for developing secure stateful CIA applications. CCF combines centralized compute with decentralized trust, supporting deployment on untrusted cloud infrastructure and transparent governance by mutually untrusted parties. CCF leverages hardware-based trusted execution environments for remotely verifiable confidentiality and code integrity. This is coupled with state machine replication backed by an auditable immutable ledger for data integrity and high availability. CCF enables each service to bring its own application logic, custom multiparty governance model, and deployment scenario, decoupling the operators of nodes from the consortium that governs them. CCF is open-source and available now at https://***/microsoft/CCF. © 2023, CC BY.

关键词： computation theory

False vacuum decay rates, more precisely

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Ai, Wen-Yuan Alexandre, Jean Sarkar, Sarben King’s College London Strand LondonWC2R 2LS United Kingdom

We develop a method for accurately calculating vacuum decay rates beyond the thin-wall regime in pure scalar field theory at the one-loop level of the effective action. It accounts for radiative effects resulting from quantum corrections to the classical bounce, including gradient effects stemming from the inhomogeneity of the bounce background. To achieve this, it is necessary to compute not only the functional determinant of the fluctuation operator in the background of the classical bounce but also its functional derivative evaluated at the classical bounce. The former is efficiently calculated using the Gel’fand-Yaglom method. We illustrate how the latter can also be calculated with the same method, combined with a computation of various Green’s functions. © 2023, CC BY.

关键词： computation theory

Desiderata for normative models of synaptic plasticity

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Bredenberg, Colin Savin, Cristina Center for Neural Science New York University New YorkNY10003 United States Mila- Quebec AI Institute 6666 Rue Saint-Urbain MontréalQCH2S 3H1 Canada Center for Data Science New York University New YorkNY10011 United States

Normative models of synaptic plasticity use a combination of mathematics and computational simulations to arrive at predictions of behavioral and network-level adaptive phenomena. In recent years, there has been an explosion of theoretical work on these models, but experimental confirmation is relatively limited. In this review, we organize work on normative plasticity models in terms of a set of desiderata which, when satisfied, are designed to guarantee that a model has a clear link between plasticity and adaptive behavior, consistency with known biological evidence about neural plasticity, and specific testable predictions. We then discuss how new models have begun to improve on these criteria and suggest avenues for further development. As prototypes, we provide detailed analyses of two specific models – REINFORCE and the Wake-Sleep algorithm. We provide a conceptual guide to help develop neural learning theories that are precise, powerful, and experimentally testable. © 2023, CC BY.

关键词： computation theory

FINITE PERIOD VECTORS AND GAUSS SUMS

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Jo, Yeongseong Department of Mathematics Education Ewha Womans University Seoul03760 Korea Republic of

We study four sums including the Jacquet-Piatetski-Shapiro-Shalika, Flicker, Bump-Friedberg, and Jacquet-Shalika sums associated to irreducible cuspidal representations of general linear groups over finite fields. By computing explicitly, we relate Asai and Bump-Friedberg gamma factors over finite fields to those over nonarchimedean local fields through level zero supercuspidal representation. Via Deligne-Kazhdan close field theory, we prove that exterior square and Bump-Friedberg gamma factors agree with corresponding Artin gamma factors of their associated tamely ramified representations through local Langlands correspondence. We also deduce product formulæ for Asai, Bump-Friedberg, and exterior square gamma factors in terms of Gauss sums. By combining these results, we examine Jacquet-Piatetski-Shapiro-Shalika, Flicker-Rallis, Jacquet-Shalika, and Friedberg-Jacquet periods and vectors and their connections to Rankin-Selberg, Asai, exterior square, and Bump-Friedberg gamma factors, respectively. © 2023, CC BY.

关键词： computation theory

Boundary triplets and the index of families of self-adjoint elliptic boundary problems

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Ivanov, Nikolai V. Department of Mathematics Michigan State University United States

The paper is devoted to an abstract axiomatic version of a construction ofboundary triplets implicit in the works of M.I. Vishik and G. Grubb and itsapplications to the index of families of self-adjoint elliptic differentialboundary problems of order one. This leads to an analytic proof of the indextheorem for Dirac-like self-adjoint boundary problems from arXiv:2207.09574,and to an Agranovich-Dynin type theorem computing the difference of indices offamilies of self-adjoint boundary problems differing only by the *** Codes 58J20, 19K56 (Primary), 47A53, 47F10 (Secondary) Copyright © 2023, The Authors. All rights reserved.

关键词： computation theory

Fundamental groups of highly symmetrical curves and Fermat line arrangments

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Amram, Meirav Roy, Praveen Kumar Sinichkin, Uriel Department of Mathematics Shamoon College of Engineering Ashdod Israel UM-DAE Centre for Excellence in Basic Sciences Mumbai India School of Mathematical Sciences Tel Aviv University Tel Aviv Israel

We showcase a computation of the fundamental group of CP2 − C when C is a curve admitting a lot of symmetries. In particular, let C denote the Fermat line arrangement in CP2 defined by the vanishing locus of homogeneous polynomial (xn−yn)(yn−zn)(zn−xn). In this article, we compute the fundamental group π1(CP2 − C) of complement of this line arrangement in the complex projective plane. We show that this group is semi-direct product of G and (formula presented), where G and Fn is defined in 4.3, and 1.2 *** Codes 14D05, 14H30, 14N20 Copyright © 2023, The Authors. All rights reserved.

关键词： computation theory