Al Hero (UM) and Jon How (MIT)
- Student collaborators: Beipeng Mu (MIT), Greg Newstadt (UM), Dennis Wei (UM).
- Interactions: Meetings between co-PI and students occurred over 2013 at conferences and other venues. In July 2013 Beipeng Mu spent 1 month at UM.
- Research goal: To adapt analytical convex optimization framework, developed by Hero'slaboratory for sensor planning in multiple target tracking problems, to
value-driven mission planning applications dveloped in How's laboratory. A paper on this work is currently in preparation for submission to ICASSP.
John Fisher (MIT) and Al Hero (UM)
- Student collaborators: Georgios Papachristoudis (MIT), Greg Newstadt (UM), Hamed Firouzi (UM).
- Interactions: In June 2013 Fisher and two of his students spent 3 days at UM.
- Research goal: Fisher and Papachristoudis have obtained distributed sensor planning strategies and characterized value of information for state prediction in the context of Gaussian observations and known topology.
The goal is to develop an analagous distrbiuted sensor planning framework for topology discovery using Poisson asymptotics developed by Firouzi and Hero.
Doug Cochran (ASU) and Al Hero (UM)
- Student collaborators: None yet.
- Interactions: In July 2013 Cochran spent 1 week at UM.
- Research goal: Cochran has been pursuing a an information geometric framework for dynamic trajectory planning.
Independently, Hero has been pursuing an information geometric framework for static classification, clustering and estimation of data that comes in the form of probability distributions. The overall goal is to combine these two perspectives.
Mike Jordan (UC Berkeley) and Al Hero (UM)
- Student collaborators: John Duchi (UC Berkeley) and Jeff Calder (UM).
- Interactions: In Feb 2013 John Duchi spent 3 days at UM. In May 2013 Jeff Calder spent 4 days at UC Berkeley.
- Research goal: Duchi and Jordan have studied the asymptotics of ranking algorithms, in particular in the context of inferring partially ordered rankings of human preferences. Calder and Hero
have studied the asymptotics of Pareto ranking, in particular for the problem of multicriterion indexing and retrieval over image databases. Our objective is to explore tangencies between these two approaches to asymptotic ranking theory and to translate them to human-in-the-loop applications like indexing and retrieval. The collaboration has impacted a publication on Pareto ranking that has been submitted by Calder and Hero to a math journal.