Workshop on Dimension Reduction and High-dimensional Inference: Theory and Applications

《降维与高维推理：理论与应用》研讨会将于2014年1月17-18日在美国佛罗里达州盖恩斯维尔的佛罗里达大学校园举行。该项目将为25名年轻研究人员和两名特邀演讲者提供旅行支持。降维领域有很长的历史，但其主要目标是以一种保留被认为相关的信息的方式对一族多变量随机向量进行降维。如今，随着高吞吐量技术和快速计算的发展，数据中的高维数据无处不在。降维现在是整个应用科学的一个流行主题，包括遗传学、食品科学、生物医学工程、经济学和计算机科学。降维领域正在迅速发展和扩大，以适应这一新的现实。在这次研讨会上，12位从事降维和高维推理工作的杰出人士将回顾该领域的现状，并介绍他们最近的工作。许多年轻的研究人员将参加研讨会，并在海报会议上展示他们的工作。降维为处理高维问题提供了一种吸引人的途径。实际上，它通过识别和组合一小部分重要变量将高维数据集转换为低维数据集，这些变量提供的信息与原始的大变量集一样多或几乎一样多。然后，可以基于低维数据集建立模型并执行估计或预测。许多现有的模型和方法不适用于高维数据，但可以应用于降维的低维数据。此外，有效的降维通常使数据可视化成为可能，这有助于后续的模型开发。降维目前是一个活跃的研究领域，但仍有许多未解决的问题。讲习班为该领域的老牌研究人员以及新来者提供了一个极好的机会，以讨论最近取得的重大进展；讨论哪些有效，哪些无效；并确定重要问题和新的研究方向。

The workshop on "Dimension Reduction and High-Dimensional Inference: Theory and Applications" will be held on January 17-18, 2014, on the campus of the University of Florida, Gainesville, FL, USA. The project will provide travel support for 25 young researchers and two invited speakers. The field of dimension reduction has a long history, but the overarching aim is to reduce the dimension of a family of multivariate random vectors in such a way that the information deemed relevant is preserved. Today, with high-throughput technologies and fast computing, high dimensionality in data is pervasive. Dimension reduction is now a prevalent theme throughout the applied sciences, including genetics, food science, biomedical engineering, economics and computer science. The area of dimension reduction is quickly evolving and expanding to adapt to this new reality. In this workshop, twelve distinguished individuals who work in dimension reduction and high-dimensional inference will review the current state of the field and present their recent work. A number of young researchers will participate in the workshop and present their work in poster sessions.Dimension reduction offers an appealing avenue for dealing with high dimensional problems. In effect it transforms a high dimensional data set to a low dimensional one by identifying and combining a small set of important variables which give as much or nearly as much information as the original large set of variables. Then one can build models and perform estimation or prediction based on the low dimensional data set. Many existing models and approaches, which do not apply to high dimensional data, can be applied to the reduced low dimensional data. In addition, effective reduction in dimension often makes it possible to visualize the data, which can facilitate subsequent model development. Dimension reduction is now an active research area, but many unsolved problems remain. The workshop provides an excellent opportunity for established researchers in the field, as well newcomers, to discuss the significant developments that have taken place recently; to discuss what works and what does not; and to identify important problems and new research directions.