BIGDATA: Collaborative Research: F: Efficient Distributed Computation of Large-Scale Graph Problems in Epidemiology and Contagion Dynamics

BIGDATA：协作研究：F：流行病学和传染动力学中大规模图问题的高效分布式计算

• 批准号: 1931628
• 负责人: Anil Kumar Vullikanti
• 金额: $ 33.08万
• 依托单位: University of Virginia Main Campus
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-02-27 至 2022-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1931628&HistoricalAwards=false
• 关键词: BIGDATA; Collaborative; Research; Efficient; Distributed

A number of phenomena of societal importance, such as the spread of diseases andcontagion processes, can be modeled by stochastic processes on networks. The analysis and control of such network phenomena involve, at their heart, fundamental graph-theoretic problems. The graphs encountered are typically of large-scale (having tens of millions of nodes); further, typical experimental analyses involve large designs with a number of parameters, leading to hundreds of thousands of graph computations. Novel methods for solving these problemsare needed, since fast response times are critical to effective decision making.The overarching goal of this project is to develop efficient distributed algorithms and associated lower bounds for graph-theoretic problems that arise in computational epidemiology and contagion dynamics. This will have a significant impact on these specific applications, through more efficient algorithmic tools for enabling complex analyses. The project will also make fundamental contributions to the design and analysis of distributed algorithms for graph problems in large-scale networks, and willresult in an algorithmic toolkit with building blocks for performing large-scale distributed graph computation. The project will lead to significant curriculum development for undergraduate as well as graduate students, as well as public health analysts. Finally, the project will help in involving minority and underrepresented students in research. The technical focus of the project will be on distributed algorithms for fundamental topics in graph algorithms such as graph connectivity, distances, subgraph analysis, and differentkinds of centrality measures. These topics underlie some of the recurring problems in the modeling, simulation and analysis and control of different kinds of contagion processes. For all these problems, the project will focus on developing provably efficient distributed algorithms and showing lower bounds under a message-passing distributed computing model. The PIs will also develop efficient implementations of these algorithms, and evaluate their performance and solution quality in real-world graphs arising in epidemiology. The graphs that arise in these applications have several novel characteristics, which will present new challenges as well as opportunities for distributed computing.

许多具有社会重要性的现象，如疾病的传播和传染过程，可以用网络上的随机过程来模拟。对这种网络现象的分析和控制，其核心是基本的图论问题。所遇到的图通常是大规模的（具有数千万个节点）;此外，典型的实验分析涉及具有许多参数的大型设计，导致数十万的图计算。解决这些问题的新方法是必要的，因为快速的响应时间是有效决策的关键。本项目的总体目标是为计算流行病学和传染动力学中出现的图论问题开发有效的分布式算法和相关的下界。 这将对这些特定应用产生重大影响，通过更有效的算法工具来实现复杂的分析。 该项目还将为大规模网络中图问题的分布式算法的设计和分析做出根本性贡献，并将产生一个算法工具包，其中包含用于执行大规模分布式图计算的构建块。 该项目将为本科生和研究生以及公共卫生分析师制定重要的课程。最后，该项目将有助于让少数民族和代表性不足的学生参与研究。该项目的技术重点将是图算法中基本主题的分布式算法，如图连通性，距离，子图分析和不同类型的中心性度量。 这些主题的基础上的一些反复出现的问题，在建模，仿真和分析和控制不同类型的传染过程。 对于所有这些问题，该项目将专注于开发可证明有效的分布式算法，并显示消息传递分布式计算模型下的下限。PI还将开发这些算法的有效实现，并在流行病学中出现的真实世界图形中评估其性能和解决方案质量。 在这些应用中出现的图有几个新的特点，这将提出新的挑战，以及分布式计算的机会。

项目成果

PREEMPT: Scalable Epidemic Interventions Using Submodular Optimization on Multi-GPU Systems

• DOI: 10.1109/sc41405.2020.00059
• 发表时间: 2020-11
• 期刊: SC20: International Conference for High Performance Computing, Networking, Storage and Analysis
• 作者: Marco Minutoli;Prathyush Sambaturu;M. Halappanavar;Antonino Tumeo;A. Kalyanaraman;A. Vullikanti

Detection of Spatiotemporal Prescription Opioid Hot Spots With Network Scan Statistics: Multistate Analysis

• DOI: 10.2196/12110
• 发表时间: 2019-04-01
• 期刊: JMIR PUBLIC HEALTH AND SURVEILLANCE
• 影响因子: 8.5
• 作者: Basak, Arinjoy;Cadena, Jose;Vullikanti, Anil
• 通讯作者: Vullikanti, Anil

Differentially Private Densest Subgraph Detection

• 发表时间: 2021-05
• 作者: Dung Nguyen;A. Vullikanti

Efficient Algorithms for Generating Provably Near-Optimal Cluster Descriptors for Explainability

• DOI: 10.1609/aaai.v34i02.5525
• 发表时间: 2020-02
• 作者: Prathyush Sambaturu;Aparna Gupta;I. Davidson;S. Ravi;A. Vullikanti;A. Warren

Controlling Epidemic Spread using Probabilistic Diffusion Models on Networks

使用网络上的概率扩散模型控制流行病传播

• 发表时间: 2022
• 期刊: Proceedings of The 25th International Conference on Artificial Intelligence and Statistics
• 作者: Babay, Amy;Dinitz, Michael;Srinivasan, Aravind;Tsepenekas, Leonidas;Vullikanti, Anil
• 通讯作者: Vullikanti, Anil