The University of Arizona

Research

Intelligent Systems

The Intelligent Systems cluster is focused on developing algorithms and processes to discover what is important and meaningful in complex data. This cluster is focused on the integration of these algorithms into information processing systems that are more intelligent, flexible and robust than currently available. This includes systems that infer representations of data that are designed for effective visualization as well as collaborative feedback from users, particularly in the case of scientific data.

Faculty

Kobus Barnard
Alon Efrat
Sandiway Fong

Projects

Computer Vision Meets Digital Libraries

Extended Description

Research in the intelligent systems cluster focuses on developing algorithms and processes to discover what is important and meaningful in complex data. We are further focused on the integration of these algorithms into information processing systems that are more intelligent, flexible and robust than currently available. This includes systems that infer representations of data that are designed for effective visualization as well as collaborative feedback from users, particularly in the case of scientific data.

The potential for impact is huge, which is reflected by the large interest in these directions by funding bodies such as NSF, ARDA, and NIH. Such systems will provide much better searching and browsing of large data sets, automatic analysis of large amounts of scientific data, and better information gathering from large sensor systems. There is real opportunity to exploit available data on a scale that is simply intractable for human analysts to go through.

These problems span many domains. Extracting relevant models and patterns in complex data applies equally to analyzing scientific data, classic problems in artificial intelligence, retrieving information from large, multi-media, data sets, and designing multi-modal information integration of sensor network data. Given the parallels of the problems, we structure this research cluster around those parallels, rather than along traditional domains. We argue that this structure promotes integrative approaches which are needed for the kinds of problems that we want to address.

Given the inter-disciplinary nature of these problems, we see foresee continuing and expanded interactions with other departments on campus and industry. We also plan to continue and expand our collaborations with the physical sciences, which increasingly requires intelligent information processing to move forward and be competitive.

This research cluster directly impacts several areas where the university has the potential to take a significant lead:

  1. innovative approaches to understanding scientific data
  2. multi-media information retrieval and mining
  3. data modeling and machine learning approaches to classic problems in intelligent systems
  4. intelligent, distributed sensor networks