Conference on Nonconvex Statistical Learning, University of Southern California, May 26-27, 2017

非凸统计学习会议，南加州大学，2017 年 5 月 26-27 日

• 批准号: 1719635
• 负责人: Jong-Shi Pang
• 金额: $ 1.5万
• 依托单位: University of Southern California
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-04-15 至 2018-03-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1719635&HistoricalAwards=false
• 关键词: Conference; Nonconvex; Statistical; Learning; University

The two-day interdisciplinary Conference on Nonconvex Statistical Learning takes place at the campus of the University of Southern California on Friday, May 26, and Saturday, May 27, 2017. The website of the conference: https://sites.google.com/a/usc.edu/cnsl2017/home will be continuously updated prior to the conference and will provide a repository for the lectures of the meeting to be made available generally. In today's digital world, huge amounts of data, i.e., big data, can be found in almost every aspect of scientific research and every walk of human activities. These data need to be managed effectively for reliable prediction, inference, and improved decision making. Statistical learning is an emergent scientific discipline wherein mathematical modeling, computational algorithms, and statistical analysis are jointly employed to address such a data management problem. The aim of the conference is to bring together researchers at all levels from multiple disciplines, including computational and applied mathematics, optimization, statistics, and engineering to report on the state of the art of the conference subject and exchange ideas for its further development. Collaborations among the participants will be fostered with the goal of advancing the science of the field of statistical learning and promoting the interfaces of the involved disciplines. The format of the conference consists of roughly two dozen lectures given by expert researchers of the field. Break times in-between the lectures are scheduled to allow discussions among all participants who will include graduate students, postdoctoral fellows, researchers in academia and industry, and faculty members in universities. This award provides support targeted for the travel expenses of junior participants. Till now, convex optimization has been a principal venue for solving many problems in statistical learning. Yet there is increasing evidence supporting the use of nonconvex formulations to enhance the realism of the models and improve their generalizations. Superior results and new advances have occurred in areas such as computational statistics, compressed sensing, imaging science, machine learning, bio-informatics and portfolio selection in which nonconvex functionals are employed to express model loss, promote sparsity, and enhance robustness. This conference provides a forum for the participants to report on their research and exchange ideas pertaining to the use of nonconvex functionals in statistical learning. The topics are organized in four main streams: modeling, advances in computation, big-data statistical learning, and innovative applications. The lectures will cover both theory and algorithms as well as promising directions for further research.

为期两天的非凸统计学习跨学科会议将于2017年5月26日(星期五)和5月27日(星期六)在南加州大学校园举行。会议网站：https://sites.google.com/a/usc.edu/cnsl2017/home将在会议前不断更新，并将提供会议讲稿的资料库，供一般人查阅。在当今的数字世界里，海量数据，即大数据，几乎存在于科学研究的方方面面和人类活动的每一步。需要对这些数据进行有效管理，以便进行可靠的预测、推断和改进决策。统计学习是一门新兴的科学学科，其中数学建模、计算算法和统计分析共同用于解决这样的数据管理问题。会议的目的是汇聚来自计算和应用数学、优化、统计学和工程学等多个学科的各级研究人员，报告会议主题的最新进展，并为其进一步发展交流意见。将促进参与者之间的合作，目标是促进统计学习领域的科学，并促进有关学科之间的联系。会议的形式由该领域的专家研究人员发表的大约24场讲座组成。讲座之间的休息时间被安排在所有参与者之间进行讨论，其中包括研究生、博士后研究员、学术界和工业界的研究人员以及大学教职员工。该奖项为初级参与者的旅费提供有针对性的支持。到目前为止，凸优化一直是解决统计学习中许多问题的主要场所。然而，越来越多的证据支持使用非凸公式来增强模型的现实性，并改进它们的概化。在计算统计、压缩传感、成像科学、机器学习、生物信息学和投资组合选择等领域，使用非凸泛函来表示模型损失、促进稀疏性和增强稳健性，都出现了优异的结果和新的进展。这次会议为与会者提供了一个论坛，报告他们关于非凸泛函在统计学习中的使用的研究和交流意见。这些主题分为四个主流：建模、计算进展、大数据统计学习和创新应用。讲座将涵盖理论和算法，以及未来研究的有希望的方向。