Jan 19, 2016 2:00 pm3 views
“Gaze-Informed Information Foraging Models for Imagery Analysis ”
Laura A. McNamara
Sandia National Laboratories Albuquerque, NM
Tuesday, January 26
12:30 – 2:00
210A Education Building
In this talk, I will discuss how and why Sandia National Laboratories employs cognitive neuroscientists, human factors psychologists, and even an anthropologist (me!) to study visual cognition, visual inspection, and information foraging problems in the RF-heavy world of Synthetic Aperture Radar. Using observations from interdisciplinary studies with intelligence groups, I’ll provide an introduction to the world of imagery analysis. These workflows are complicated amalgamations of visual inspection and information foraging behaviors, supported by a wide range of image products, viewing platforms, and tools. Of particular importance is the shift over the past couple of decades from so-called “hardcopy” to “softcopy” image analysis workflows: not only have the tools and techniques of image analysis changed, but analysts can access an increasingly diverse set of highly specialized image types. As any of our team members can attest, expert imagery analysts are extremely good – and very fast – in detecting, evaluating, and extracting meaningful signatures from very large, diverse sets of imagery
Over the past few years, we’ve come to appreciate the potential for eye tracking data to help us understand the chains of micro-decisions that describe an imagery analyst’s path through a geospatial information space. The ability to associate gaze events with image features in dynamic, user-driven workflows could reveal how imagery analysts acquire the skills necessary to extract information efficiently and accurately from geospatial datasets. In practice, however, it is extremely difficult to study gaze-contingent decision-making in realistic, user-driven workflows. Therefore, we’ve recently embarked on a methodological/software development project to create tools that will enable us to integrate gaze data with complementary behavioral indicators of analytic decision-making. Ultimately, we’d like to enable researchers to establish a theoretically sound, well-characterized empirical foundation for the design of visual analytic systems, workflows, and training protocols.
For questions about this brownbag, please contact Professor Liz Stine-Morrow at eals@illinois.edu
