Interactive visual analytics for high dimensional data

Abstract

Many modern data sets can be represented in high dimensional vector spaces and have benefited from computational methods that utilize advanced techniques from numerical linear algebra and optimization. Visual analytics approaches have contributed greatly to data understanding and analysis due to utilization of both automated algorithms and human's quick visual perception and interaction. However, visual analytics targeting high dimensional large-scale data has been challenging due to low dimensional screen space with limited pixels to represent data. Among various computational techniques supporting visual analytics, dimension reduction and clustering have played essential roles by reducing the dimension and volume to visually manageable scales. In this talk, we present some of the key foundational methods for supervised dimension reduction such as linear discriminant analysis (LDA), dimension reduction and clustering/topic discovery by nonnegative matrix factorization (NMF), and visual spatial alignment for effective fusion and comparisons by Orthogonal Procrustes. We demonstrate how these methods can effectively support interactive visual analytic tasks that involve large-scale document and image data sets.