Research Areas
Artificial intelligence, machine learning, automated planning and reasoning, natural languageprocessing.
Research Description
My main research interest is to understand learning and thinking by simulating them in computers. My work ranges from theoretical analyses of learning problems and algorithms to their implementation, evaluation, and application to real-world problems.
One of my research thrusts is to learn to act intelligently by building models of actions, planning or reasoning with them, executing the plans, and learning better models and ways to plan with them. There are several interesting questions here such as how to model only what is needed, how to account for errors in the models, how best to combine knowledge and search in planning, how to learn from observation as well as practice, how to exploit hierarchies in planning and learning, and how to transfer knowledge from one domain to a related domain.
Another research thrust is to learn in structured contexts such as natural language processing where the examples have rich internal structure, and are noisy, incomplete, biased, and highly interconnected.
The interesting issues here include learning in expressive representations, learning multiple related concepts simultaneously, constraining learning by explicit prior knowledge, accounting for bias and incompleteness in the data, and integrated learning and reasoning.
Applications of Research
The applications of research include fire and medical emergency response, learning and training systems for expert tasks, intelligent assistants for disabled and elderly, and personalized medicine.