Research

Thomas R. Ioerger

Research Interests

My primary research interests are in the areas of Artificial Intelligence and Machine Learning. In Machine Learning, my main focus is on developing automated methods for improving the accuracy of learning algorithms through enhancing the representation of examples. This includes new algorithms that make learning algorithms more accurate by identifying the most relevant features (feature weighting), discovering complex relationships among features (feature interactions), and creating new features through mathematical operations (feature extraction/construction). This has applications for improving pattern recognition or data mining in a wide variety of domains.

A second area of interest for me is in Intelligent Agents. In the past, I worked on modeling of user interests in documents from relevance feedback, and on modeling emotions in believable agents. More recently, in collaboration with Dr. John Yen and Dr. Richard Volz in this department, we have been developing agent-based techniques for building intelligent team-training systems. This requires the ability to simulate the complex reasoning that team members must do about the roles, responsibilities, and beliefs of their teammates. We have developed novel teamwork algorithms for simulating collaborative behavior in virtual team-member agents, and user-modeling approaches for understanding the complex decision-making involved in trainees' actions within a team.

I also work in the area of Bioinformatics (or Computational Biology). Within this area, my current interests are in: 1) improving the accuracy of sequence alignment algorithms, 2) molecular modeling of protein structures, and 3) automated interpretation of electron density maps in X-ray crystallography. My overall approach is to combine my research interests via interdisciplinary projects, in which machine learning techniques are brought to bear on difficult problems in biology, and conversely, these challenging problems are used to motivate the development of new techniques in machine learning, especially change-of-representation methods.

Research Support

Automated X-ray Crystallography for Structural Genomics Automated Intelligent Group and Team Training Systems Automated Identification of Flow Patterns in Congested Traffic Digitization Research in Support of Force XXI Pattern Recognition in Macromolecular Crystallography A Machine Learning Approach to Modeling Immunoglobulins Computational Resources for Training and Research in the Areas of Computational Biology and Bioinformatics Development of Command and Control (C2) Collaborative Agents for Simulating Teamwork for the Research and Development Centers (RDEC) Federation

Brief Summary of My Research Interests (1 PostScript slide)

What is Data-Mining?

A Program for Identifying Cys-Cys-Trp Triad Motifs in Protein Structures

JARE - Java Automated Reasoning Engine

Journals

Artificial Intelligence Protein Science

Machine Learning Links

Molecular Biology Links

Other Research Links