Collaborative Research: SHF: Medium: NetSplicer: Scalable Decoupling-based Algorithms for Multilayer Network Analysis

合作研究：SHF：中：NetSplicer：用于多层网络分析的可扩展的基于解耦的算法

• 批准号: 1955971
• 负责人: Kamesh Madduri
• 金额: $ 30.2万
• 依托单位: Pennsylvania State Univ University Park
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-07-01 至 2024-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1955971&HistoricalAwards=false
• 关键词: Collaborative; Research; SHF; Medium; NetSplicer

A multilayer network is a powerful and expressive mathematical tool for modeling and analyzing social, economic, biological, and technological systems. Informally, a multilayer network is a collection of related graphs. Applications of multilayer networks include understanding social networks, economic systems, online marketplaces, and detecting vulnerabilities in cyber-physical systems. While this research area is rapidly growing, there is a dearth of computational tools for analyzing large-scale networks from diverse applications. This project will develop the theoretical foundations and software infrastructure for analyzing very large multilayer networks on modern computing systems, thereby enabling their widespread use in diverse applications.This project will develop NetSplicer, a collection of scalable high-performance algorithms for multilayer-network analysis. The approaches in NetSplicer will be based on a divide-and-conquer-like technique called network decoupling. Using decoupling, the multilayer network can be subdivided into multiple components, each of which could be potentially analyzed using known graph algorithms. Network decoupling seeks to address issues that are critical for multilayer analysis, such as reducing information loss and preserving structural and semantic information. The challenges in efficiently applying network decoupling include determining optimal decoupling strategies, preserving the structure and content of multilayer networks that have multiple vertex and edge types, and developing architecture-aware scalable algorithms that apply across different layers of a network. This project will provide a new capability for multiple research communities and will build a repository for multilayer networks. The planned collaborations with domain scientists from academia and industry, as well as curriculum development and outreach activities, will shape project development efforts to maximize impact.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

多层网络是一种强大的数学工具，用于建模和分析社会，经济，生物和技术系统。非正式地，多层网络是相关图的集合。多层网络的应用包括理解社交网络、经济系统、在线市场和检测网络物理系统中的漏洞。虽然这一研究领域正在迅速发展，但缺乏用于分析来自不同应用的大规模网络的计算工具。本项目将开发在现代计算系统上分析超大型多层网络的理论基础和软件基础设施，从而使其在各种应用中得到广泛应用。本项目将开发用于多层网络分析的可扩展的高性能算法集合NetSplicer。NetSplicer中的方法将基于一种称为网络解耦的分而治之的技术。使用解耦，多层网络可以被细分为多个组件，每个组件都可以使用已知的图形算法进行分析。网络解耦旨在解决对多层分析至关重要的问题，例如减少信息丢失以及保留结构和语义信息。有效应用网络解耦的挑战包括确定最佳解耦策略，保留具有多个顶点和边类型的多层网络的结构和内容，以及开发适用于网络不同层的架构感知可扩展算法。该项目将为多个研究社区提供新的能力，并将为多层网络建立一个存储库。计划与学术界和工业界的领域科学家合作，以及课程开发和推广活动，将塑造项目开发工作，以最大限度地发挥影响力。该奖项反映了NSF的法定使命，并被认为值得通过使用基金会的智力价值和更广泛的影响审查标准进行评估来支持。

项目成果

Performance-Portable Graph Coarsening for Efficient Multilevel Graph Analysis

高性能便携式图形粗化，用于高效的多级图形分析

• DOI: 10.1109/ipdps49936.2021.00030
• 发表时间: 2021
• 期刊: 2021 IEEE International Parallel and Distributed Processing Symposium (IPDPS
• 作者: Gilbert, Michael S.;Acer, Seher;Boman, Erik G.;Madduri, Kamesh;Rajamanickam, Sivasankaran
• 通讯作者: Rajamanickam, Sivasankaran