A New Multidisciplinary Approach to Data Understanding: Integrating Human and Computational Approaches
The goal of this seminar is to begin laying the foundation for a new field that explicitly unites human and algorithmic approaches to pattern recognition. Five million years of human evolution have made us exquisitely capable of seeing patterns and features in the natural world. Our agenda is to understand better and to couple these natural capabilities with the tools of modern computer-based systems. Isolated point solutions have been proposed in the literature, which explicitly combine human and computer capabilities, but not a cross-disciplinary, general framework, which limits their impact and effectiveness.
This seminar defines the first step in creating a general research agenda and focuses on 1) how human observers explore, interact, and navigate through visual representations of data and models; 2) how algorithms from machine learning, computer vision, visualization, and computer graphics can help scientists (e.g., astronomers, biologists, doctors, etc.) extract features from data; and 3) how to capture these features mathematically so that they can be used for comparison and search. The purpose of the seminar is to assemble scientific leaders in imaging, computer vision, visualization, perception, and design, along with scientific leaders representing a wide range of disciplines and approaches in this space, to help forge this new synergy. Working together, we will produce a set of coherent questions, approaches, and challenges that can be used to build a research roadmap to shape the future of this discipline, which can be leveraged for large–scale grant development.