Collaborative Research: Risk-Averse Cluster Detection in Network Models of Bigdata Under Measurement Uncertainty

合作研究：测量不确定性下大数据网络模型中的风险规避聚类检测

• 批准号: 1404971
• 负责人: Balabhaskar Balasundaram
• 金额: $ 27.16万
• 依托单位: Oklahoma State University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2014
• 资助国家: 美国
• 起止时间: 2014-04-15 至 2018-03-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1404971&HistoricalAwards=false
• 关键词: Collaborative; Research; Risk; Averse; Cluster

The objective of this collaborative research award is to establish a comprehensive knowledge-base for detecting low-diameter clusters in network models of social and biological big-data that are subject to measurement errors and incomplete information. This research study focuses on a novel model for detecting clusters in social and biological networks called a "k-club." Some key innovations pursued as part of this study include the use of a risk measure from financial engineering that is used to quantify losses in cluster cohesiveness that results from the measurement errors and uncertainty that underlies such network models of big data. This measure represents the downside risk of high losses in the worst-case scenarios leading to risk-averse mathematical models that facilitate the detection of clusters that are more likely to be application significant. The models and algorithms that result from this study will be validated using laboratory experimental data and expertise provided by collaborators in biological sciences, and also by using publicly available social and financial network datasets.If successful, this work can enable biologists to better understand complex biological networks, accelerate discoveries and reduce experimental costs. The research has the potential to spur fundamental discoveries that can lead to advances in biomedicine, agriculture and bioenergy with positive societal impacts. High School mathematics teachers will be engaged to receive first-hand experience in large-scale network/data analysis as well as identify and transfer appropriate material to their classroom via active learning exercises. Enhancing minority participation and professional development are key outreach goals of this project. Graduate and undergraduate students from under-represented groups will be actively sought to become engaged in the research and hopefully the professoriate as recommended by the National Science Board and aligned with the America COMPETES Reauthorization Act.

该协作研究奖的目的是建立一个全面的知识库，用于在社会和生物大数据的网络模型中检测低直径群集，该集群受到测量错误和不完整的信息。这项研究的重点是一种新的模型，用于检测称为“ K-Club”的社会和生物网络中的聚类。作为这项研究的一部分，一些关键的创新包括使用金融工程的风险措施，用于量化集群凝聚力中的损失，这是由于测量错误和不确定性是基于此类大数据网络模型的基础的。该措施代表了最严重的情况下高损失的下行风险，从而导致规避风险的数学模型，这些模型有助于检测更可能具有应用的群集。 这项研究产生的模型和算法将通过生物科学合作者提供的实验室实验数据和专业知识以及使用公开可用的社会和金融网络数据集进行验证。如果成功，这项工作可以使生物学家能够更好地了解复杂的生物网络，加速发现并降低实验成本。这项研究有可能刺激基本发现，这可能会导致生物医​​学，农业和生物能源的进步，并产生积极的社会影响。高中数学教师将参与大规模网络/数据分析中的第一手经验，并通过积极的学习练习来识别和将适当的材料转移到其课堂上。加强少数族裔参与和专业发展是该项目的关键外展目标。来自代表性不足的团体的研究生和本科生将积极寻求参与研究，并希望按照国家科学委员会的建议并与《美国竞争重新授权法》保持一致。