webb@webb.applicon.UUCP (10/21/88)
I recently posted a request for information about graduate schools which have good programs in Artificial Intelligence and Machine Learning. This is a summary of the information which I recieved. To all those who responded, thank you very much. I invite further comments on the opinions expressed below, and further input from those at these or other schools. ******Eastern Schools: Rutgers: - Strong learning program. University of North Carolina: - No AI program. Yale: - Dominated by Roger Schank, who is reputedly very hard on his students. Strong recommendations against going here. - Dana Angluin doing excellent theoretical work. Harvard: - Small program (5-6 students/year), correspondingly close contact with faculty. - Les Valiant is doing theoretical machine learning work. - William Woods is willing to support machine learning work, though his usual field is natural language. Carniege-Mellon University: - Very difficult to get in. - Rated consistiently as one of the top AI and Machine Learning schools in the world. - Diverse program - Allen Newell; SOAR project - Tom Mitchell, Jamie Carbonell, John Anderson in Machine Learning, many others in other fields of AI and connectionism. Berliner, Kenade, Reddy, Hinton, etc. - Focus on research rather than classwork. University of Pennsylvania: - Well-known for their natural language work, not so much so for machine learning. - One complaint about terrible student/administration relationships. MIT: - Very difficult to get in. - Famous for requiring 8-9 years of work for PhD. - Rumored: (from Stanford student) - Unfriendly - One dimensional Department. - Many professors were MIT undergrads. University of Mass. @ Amherst: - Strong AI and learning programs. Georgia Tech: - Dr. Janet Kolodner; Case Based Reasoning, Experiential learning, PhD from Yale under Roger Schank. - Connection with DARPA through Col. Bob Simpson who recieve MS in Machine Learning from Georgia Tech under Kolodner. He is head of DARPA Machine Learning research. University of Pittsburgh: - Bruce Buchanan has come here from Stanford to set up a big-time AI lab. If he stays, excitement will follow. - Focus on Expert Systems. ******Central Schools: University of Illinois @ Champaign-Urbana: - 6 AI faculty whose primary interest is learning, 4 have it as a secondary interest. Fields include: - EBL (Jerry DeJong) - Theory of Learning (Lenny Pitt) - Probabalistic learning, applied and theoretical (Sylvian Ray, Larry Rendell) - Conceptual Clustering (Bob Stepp) - KBS Learning, automated programming (David Wilkins) - Interdiscplinary approach, esp. re. the psychology dept. - Doug Medin, Dedre Genter, William Brewer, William Greenough - Work also being done in Lingusitics, Statistics, Electrical Engineering and Physics Depts. - Beckman Institute on campus - Brand new $50M facility for study of intelligence and complex systems. University of Michigan: - Holland; Classifiers and Genetic Algorithms - Host of last year's (1987) Machine Learning conference. ******Western Schools: University of Texas @ Austin: - Machine Learning group headed by Bruce Porter. - Many well-known and respected scientists working and visiting there. (eg. Silberschatz, Boyer and Moore, Dijkstra) - Relationship with MCC and Doug Lenat. Stanford: - Very difficult to get in. - Famous for requiring 8-9 years of work for PhD. - Bruce Buchanan, their best learning professor, has relocated to U. Pittsburgh. - AI department is dominated by those who believe that rigorous logic is the representation best suited to solving problems. - Rich Keller; explaination based learning. Their only specialist. - David Rumelhart, connectionist, works in psych. dept. - Most professors will support machine learning research however. - Terrific connections with industry: - Schlumberger - NASA Ames - Xerox PARC - Lockheed AI Center - Do not have an active learning group. University of California @ San Diego: - Most, if not all, of their machine learning work is centered around connectionism. University of California @ Berkeley: - AI is not the focus of their CS department. - Main AI professor is Wilensky, a clone of Roger Schank. - Stuart Russell, Stanford graduate. University of California @ Irvine: - Strong psychological orientation. - Good funding, good equipment. - CS dept. is up and coming. - Pat Langely main Learning professor. - 4 faculty doing learning work - 2 doing explaination-based learning - 1 doing empirical work - 45min to 1hr from LA. University of California @ Los Angeles: - Not recommended for machine learning ******Foreign schools: University of Edinburgh, Scotland: Peter Webb. {allegra|decvax|harvard|yale|mirror}!ima!applicon!webb, {mit-eddie|raybed2|spar|ulowell|sun}!applicon!webb, webb@applicon.com