Aravind K. Joshi
Professor of Computer and Cognitive Science
Henry Salvatori Professor of Linguistics, and
Co-director of the Institute for Research in Cognitive Science
at the University of Pennsylvania, in Philadephia, Pennsylvania, USA.
Relationship Between Natural Language Processing and AIThe use of constrained formal/computational systems just adequate for modeling various aspects of language - syntax, semantics, pragmatics and discourse, among others has proved to be not only an effective research strategy but has led to deeper understanding of these aspects, with implications to both machine processing as well as human processing. This approach enables one to distinguish between universal and stipulative constraints. The other approach is to start with the most general and most powerful formal/computational system and use it to model these phenomena, thus making all constraints stipulative, in a sense. The tension between these approaches, together with the increasing use of empirical methods combining structural and statistical information, has made the relationship between natural language processing and AI quite stormy. I will review some of the past and current research following the first approach and also suggest how the stormy relationship could be improved.
Leslie P. Kaelbling
Associate Professor of Computer Science at Brown University, in Providence, Rhode Island, USA
She is the author of one book, "Learning in Embedded Systems" and the editor of another. She is the recipient of the National Science Foundation National Young Investigator Award and of the National Science Foundation Presidential Faculty Award."
Why Robbie Can't Learn: The Difficulty of Learning in Autonomous AgentsIn recent years, machine learning methods have enjoyed great success in a variety of applications. Unfortunately, on-line learning in autonomous agents has not generally been one of them. Reinforcement-learning methods that were developed to address problems of learning agents have been most successful in off-line applications. In this talk, I will briefly review the basic methods of reinforcement learning, point out some of their shortcomings, argue that we are expecting too much from such methods, and speculate about how to build complex, adaptive autonomous agents.
The Hong Kong University of Science and Technology
Jaime Carbonell, Yiming Yang, Robert E. Frederking, Ralf Brown, Yibing Geng, and Daniel Lee
Carnegie Mellon University
Timothy Huang and Stuart Russell
University of California at Berkeley