检索结果-内蒙古大学图书馆

33rd International symposium on High-Performance Parallel and distributed computing (HPDC)

作者： Joaquim, Molo Mbasa Univ Pisa Pisa Italy Natl Res Council Italy Pisa Italy

ISBN: (纸本)9798400704130

Recent years have witnessed significant advancements in Artificial Intelligence (AI), particularly with the rise of Deep Neural Networks fueled by large datasets and increased model complexity. However, the demand for substantial computational resources poses challenges in centralized data scenarios. Edge Intelligence (EI), combining Edge computing and AI, emerges as a transformative solution for decentralized learning, crucial in the era of IoT proliferation. While Federated Learning (FL) has been a prominent paradigm in decentralized learning, its limitations have prompted researchers to explore alternative solutions using Knowledge Distillation (KD) as a basis. The purpose of this Ph.D. research is to explore KD as a new paradigm for decentralized learning, contribute to enhancing its performance, and study the trade-off between FL and KD in terms of efficiency and effectiveness to identify best practices and insights in EI environments.

关键词： Knowledge Distillation Computer Vision Object Recognition Deep Learning Edge Intelligence

来源：评论

学校读者我要写书评

暂无评论

Teaching Ethics & Activism in a Human-Computer Interaction Professional Master's Program 2024

Teaching Ethics & Activism in a Human-Computer Interaction P...

引用

55th acm Technical symposium on Computer Science Education (SIGCSE)

作者： Rivera, Veronica A. Su, Norman Makoto Stanford Univ Stanford CA USA Univ Calif Santa Cruz Santa Cruz CA USA

ISBN: (纸本)9798400704239

We report on a new ethics course for industry-bound students in a Human-Computer Interaction (HCI) professional master's program. The goal is to prepare students to think critically about the technology they design and to drive ethical change within their future organizations. Unlike research-oriented graduate programs, students in professional master's programs primarily seek to enter industry, oftentimes making a career change from a non-computing area. Thus, in addition to supporting students' ethical reasoning skills, ethics pedagogy needs to also help students develop core skills in the technical discipline and be grounded in real-world situations to align with their practice-based goals. To achieve this, we structure our course around three principles: survey (introducing contemporary ethical issues in computing and their social context), stakeholders (considering the multi-faceted nature of ethical decision-making), and skills (developing the technical and communication skills needed to drive ethical change). We hope that our curriculum and reflections will help other instructors connect ethics pedagogy to professional practice in the classroom.

关键词： ethics professional training HCI UX design project-based learning computing education

来源：评论

学校读者我要写书评

暂无评论

Revisiting the Power of Non-Equivocation in distributed Protocols 22

Revisiting the Power of Non-Equivocation in Distributed Prot...

引用

acm symposium on principles of distributed computing (PODC)

作者： Ben-David, Naama Chan, Benjamin Y. Shi, Elaine VMware Res Palo Alto CA 94304 USA Cornell Univ Ithaca NY USA CMU Pittsburgh PA USA

ISBN: (纸本)9781450392624

Trusted hardware and new computing platforms such as RDMA naturally provide a non-equivocation abstraction. Previous works have shown that non-equivocation allows us to achieve tasks that otherwise would not have been possible in the plain model. In this paper, we are interested in understanding whether we can use nonequivocation to compile any asynchronous crash-fault protocol into one that tolerates the same number of Byzantine faults. Furthermore, we consider protocols with security and privacy guarantees that we must preserve under the compilation. Previous works have aimed to achieve a similar goal. However, we explain why the previous results in this area were incomplete. We then present a new compiler that achieves security and privacy, and does so while introducing only polynomial overhead over the underlying protocol (as compared to exponential overhead in previous results).

关键词： non-equivocation multiparty computation Byzantine fault tolerance

来源：评论

学校读者我要写书评

暂无评论

MAD-Max Beyond Single-Node: Enabling Large Machine Learning Model Acceleration on distributed Systems 51

MAD-Max Beyond Single-Node: Enabling Large Machine Learning ...

引用

acm/IEEE 51st Annual International symposium on Computer Architecture (ISCA)

作者： Hsia, Samuel Golden, Alicia Acun, Bilge Ardalani, Newsha DeVito, Zachary Wei, Gu-Yeon Brooks, David Wu, Carole-Jean FAIR Meta Menlo Pk CA 94025 USA Harvard Univ Cambridge MA 02138 USA

ISBN: (纸本)9798350326598;9798350326581

Training and deploying large-scale machine learning models is time-consuming, requires significant distributed computing infrastructures, and incurs high operational costs. Our analysis, grounded in real-world large model training on datacenter-scale infrastructures, reveals that 14 similar to 32% of all GPU hours are spent on communication with no overlapping computation. To minimize this outstanding communication latency and other inherent at-scale inefficiencies, we introduce an agile performance modeling framework, MAD-Max. This framework is designed to optimize parallelization strategies and facilitate hardware-software co-design opportunities. Through the application of MAD-Max to a suite of real-world large-scale ML models on state-of-the-art GPU clusters, we showcase potential throughput enhancements of up to 2.24x for pre-training and up to 5.27x for inference scenarios, respectively.

关键词： Data Center distributed Inference distributed Training GPU Hardware-Software Co-Design Machine Learning Parallelization Performance Model Simulator

来源：评论

学校读者我要写书评

暂无评论

Pure: Evolving Message Passing To Better Leverage Shared Memory Within Nodes 24

Pure: Evolving Message Passing To Better Leverage Shared Mem...

引用

29th acm SIGPLAN Annual symposium on principles and Practice of Parallel Programming (PPoPP)

作者： Psota, James Solar-Lezama, Armando MIT CSAIL Cambridge MA 02139 USA

ISBN: (纸本)9798400704352

Pure is a new programming model and runtime system explicitly designed to take advantage of shared memory within nodes in the context of a mostly message passing interface enhanced with the ability to use tasks to make use of idle cores. Pure leverages shared memory in two ways: (a) by allowing cores to steal work from each other while waiting on messages to arrive, and, (b) by leveraging *** lock-free data structures in shared memory to achieve highperformance messaging and collective operations between the ranks within nodes. We use microbenchmarks to evaluate Pure's key messaging and collective features and also show application speedups up to 2.1 Chi on the CoMD molecular dynamics and the miniAMR adaptive mesh *** applications scaling up to 4,096 cores.

关键词： parallel programming models distributed runtime systems task-based parallelism concurrent data structures lock-free data structures

来源：评论

学校读者我要写书评

暂无评论

A distributed algorithm for directed minimum-weight spanning tree

引用

distributed computing 2023年第1期36卷 57-87页

作者： Fischer, Orr Oshman, Rotem Tel Aviv Univ Comp Sci Dept Tel Aviv Israel

In the directed minimum spanning tree problem (DMST, also called minimum weight arborescence), we are given a directed weighted graph, and a root node r. Our goal is to construct a minimum-weight directed spanning tree, rooted at r and oriented outwards. We present the first sub-quadraticDMSTalgorithm in the distributed CONGEST network model, where the messages exchanged between the network nodes are bounded in size. We consider three versions of the model: a network where the communication links are bidirectional but can have different weights in the two directions;a network where communication is unidirectional;and the Congested Clique model, where all nodes can communicate directly with each other. Our DMST algorithm is based on a variant of Lovasz' DMST algorithm for the PRAM model, and uses a distributed single-source shortest-path (SSSP) algorithm for directed graphs as a black box. In the bidirectional CONGEST model, our algorithm has roughly the same running time as the SSSP algorithm that is used as a black box;using the state-of-the-art SSSP algorithm due to Chechik and Mukhtar (in: symposium on principles of distributed computing (PODC), acm, 2020, pp 464-473), we obtain a running time of (O) over tilde(root nD(1/4) + D)) rounds for the bidirectional communication case. For the unidirectional communication model we give an (O) over tilde (n) algorithm, and show that it is nearly optimal. And finally, for the Congested Clique, our algorithm again matches the best known SSSP algorithm: it runs in (O) over tilde (n(1/3)) rounds. On the negative side, we adapt an observation of Chechik in the sequential setting to show that in all three models, the DMST problem is at least as hard as the (s, t)-shortest path problem. Thus, in terms of round complexity, distributed DMST lies between single-source shortest path and (s, t)-shortest path.

关键词： distributed computing Directed minimum spanning tree Arborescence CONGEST

来源：评论

学校读者我要写书评

暂无评论

Byzantine Resilient distributed computing on External Data 38

Byzantine Resilient Distributed Computing on External Data

引用

38th International symposium on distributed computing, DISC 2024

作者： Augustine, John Biju, Jeffin Meir, Shachar Peleg, David Ramachandran, Srikkanth Thiruvengadam, Aishwarya Indian Institute of Technology Madras Chennai India Weizmann Institute of Science Rehovot Israel

ISBN: (纸本)9783959773522

We study a class of problems we call retrieval problems in which a distributed network has read-only access to a trusted external data source through queries, and each peer is required to output some computable function of the data. To formalize this, we propose the Data Retrieval Model comprising two parts: (1) a congested clique network with k peers, up to βk of which can be Byzantine in every execution (for suitable values of β ∈ [0, 1));(2) a trusted source of data with no computational abilities, called the External Data Source (or just source for short). This source stores an array X of n bits (n k), providing every peer in the congested clique read-only access to X through queries. It is assumed that a query to the source is significantly more expensive than a message between two peers in the network. Hence, we prioritize minimizing the number of queries a peer performs over the number of messages it sends. Retrieval problems are easily solved by having each peer query all of X, so we focus on designing non-trivial query-efficient protocols for retrieval problems in the DR network that achieve low query performance per peer. Specifically, to initiate this study, we present deterministic and randomized upper and lower bounds for two fundamental problems. The first is the Download problem that requires every peer to output an array of n bits identical to X. The second problem of focus, Disjunction, requires nodes to learn if some bit in X is set to 1. © John Augustine, Jeffin Biju, Shachar Meir, David Peleg, Srikkanth Ramachandran, and Aishwarya Thiruvengadam.

关键词： Spatio-temporal data

来源：评论

学校读者我要写书评

暂无评论

Power-over-Skin: Full-Body Wearables Powered By Intra-Body RF Energy 24

Power-over-Skin: Full-Body Wearables Powered By Intra-Body R...

引用

37th acm Annual symposium on User Interface Software and Technology (UIST)

作者： Kong, Andy Kim, Daehwa Harrison, Chris Carnegie Mellon Univ Pittsburgh PA 15213 USA

ISBN: (纸本)9798400706288

Powerful computing devices are now small enough to be easily worn on the body. However, batteries pose a major design and user experience obstacle, adding weight and volume, and generally requiring periodic device removal and recharging. In response, we developed Power-over-Skin, an approach using the human body itself to deliver power to many distributed, battery-free, worn devices. We demonstrate power delivery from on-body distances as far as from head-to-toe, with sufficient energy to power microcontrollers capable of sensing and wireless communication. We share results from a study campaign that informed our implementation, as well as experiments that validate our final system. We conclude with several demonstration devices, ranging from input controllers to longitudinal bio-sensors, which highlight the efficacy and potential of our approach.

关键词： Intra-body power transfer IBPT wireless power on-body devices wearables digital jewelry eTattoos sensing ubiquitous computing

来源：评论

学校读者我要写书评

暂无评论

HM2: Efficient Host Memory Management for RDMA-Enabled distributed Systems 23

HM2: Efficient Host Memory Management for RDMA-Enabled Distr...

引用

32nd International symposium on High-Performance Parallel and distributed computing (HPDC) part of the acm Federated computing Research Conference (FCRC)

作者： Tang, Chen Chu, Zhaole Jin, Peiquan Luo, Yongping Guo, Kuankuan Univ Sci Technol China Hefei Peoples R China ByteDance Inc Beijing Peoples R China

ISBN: (纸本)9798400701559

Remote direct memory access (RDMA) supports zero-copy networking by transferring data from clients directly to host memory, eliminating the need to copy data between clients' memory and the data buffers in the hosting server. However, the hosting server must design efficient memory management schemes to handle incoming clients' data. In this paper, we propose a high-performance host memory management scheme called HM2 for RDMA-enabled distributed systems. We present a new buffer structure for incoming data from clients. In addition, we propose efficient data processing methods to reduce network transfers between clients and servers. We conducted a preliminary experiment to evaluate HM2, and the results showed HM2 achieved higher throughput than existing schemes, including L5 and FaRM.

关键词： RDMA memory management data buffer

来源：评论

学校读者我要写书评

暂无评论

Optimal Communication Bounds for Classic Functions in the Coordinator Model and Beyond 2024

Optimal Communication Bounds for Classic Functions in the Co...

引用

56th Annual acm symposium on Theory of computing (STOC)

作者： Esfandiari, Hossein Kacham, Praneeth Mirrokni, Vahab Woodruff, David P. Zhong, Peilin Google London England Carnegie Mellon Univ Pittsburgh PA USA Google Res New York NY USA

ISBN: (纸本)9798400703836

In the coordinator model of communication with B servers, given an arbitrary non-negative function f, we study the problem of approximating the sum Sigma i([n]) f(x(i)) up to a 1 +/- epsilon factor. Here the vector x is an element of R-n is defined to be x = x ( 1)+ center dot center dot center dot+x(s), where x(j) >= 0 denotes the non-negative vector held by the 9 -th server. A special case of the problem is when f(x) = x(k) which corresponds to the well-studied problem of F-k moment estimation in the distributed communication model. We introduce a new parameter c(f)[s] which captures the communication complexity of approximating Sigma(i is an element of[n]) f(x(i)) and for a broad class of functions f which includes f(x) = x(k) for k >= 2 and other robust functions such as the Huber loss function, we give a two round protocol that uses total communication c(f)[s]/epsilon(2) bits, up to polylogarithmic factors. For this broad class of functions, our result improves upon the communication bounds achieved by Kannan, Vempala, andWoodruff (COLT 2014) andWoodruff and Zhang (STOC 2012), obtaining the optimal communication up to polylogarithmic factors in the minimum number of rounds. We show that our protocol can also be used for approximating higher-order correlations. Our results are part of a broad framework for optimally sampling from a joint distribution in terms of the marginal distributions held on individual servers. Apart from the coordinator model, algorithms for other graph topologies in which each node is a server have been extensively studied. We argue that directly lifting protocols from the coordinator model to other graph topologies will require some nodes in the graph to send a lot of communication. Hence, a natural question is the type of problems that can be efficiently solved in general graph topologies. We address this question by giving communication efficient protocols in the so-called personalized CONGEST model for solving linear regression and lo

关键词： distributed algorithms coordinator model moment estimation personalized CONGEST model

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：