Otzenhausen, Germany, October 8--10, 1998.
The 9th International Conference on Algorithmic Learning Theory (ALT) was held in Otzenhausen, Germany. This is the ninth in a series of annual conferences started in 1990 and it was supported jointly by the Japanese Society for Artificial Intelligence (JSAI) and the University of Kaiserslautern.
For the technical sessions, 52 authors contributed 3 invited and 26 (out of 34 submitted) selected talks to the program. The contributions ranged over the following subjects: inductive logic programming, data mining, inductive inference, learning via queries, prediction algorithms, learning formal languages, complexity of neural trees, polynomial time learning from positive data. The three invited talks where presented by distinguished speakers, who gave nice and challenging overviews on their research fields and working areas:
(1) Stefan Wrobel (GMD, Sankt Augustin, Germany) presented work on ``Scalability issues in inductive logic programming'': If stated in an unrestricted fashion, the inductive acquisition of first-order predicative theories is undecidable. Nevertheless, several new approaches to inductive logic programming have useful and efficient applications within the field of data mining or knowledge discovery in databases. The author gave an overview of the basic aspects of scalability in inductive logic programming and introduced to the system MIDOS, a recent first-order subgroup discovery algorithm.
(2) Akira Maruoka (Tohoku University, Sendai, Japan) presented joint work with Eiji Takimoto on ``Structured Weight-based prediction algorithms'': The talk focussed on design methods for a variety of on-line prediction algorithms based on the structured weight-based prediction algorithms introduced by Takimoto, Maruoka and Vovk. The typical expert model is generalized to the case where the experts are laid on a tree structure and not on one layer. This expert model can be applied to search for the best pruning in a straightforward fashion through dynamic programming scheme.
(3) Arun Sharma (The University of New South Wales, Australia) presented joint work with Eric McCreath on ``LIME: A system for learning relations'': LIME is a system that employs a Bayesian heuristic to evaluate logic programs as hypotheses. It deals in particular with the notion of simple clauses. The learning is implemented such that LIME combines simple clauses to construct a set of candidate clauses. Subsets of these candidate clauses are then evaluated via the Bayesian heuristic to find the final hypothesis. LIME verified in experiments that it meets the following design-goals: it has an improved noise handling, learns from fixed sets of positive examples and infers recursive logic programs. At the end of his talk, the author compared LIME with the systems FOIL and PROGOL.
The conference was chaired by Professor Michael M. Richter (Universität Kaiserslautern) and Professor Carl H. Smith (University of Maryland, co-chair); Professor Rolf Wiehagen (Universität Kaiserslautern) served as Program Committee* Chair and Professor Thomas Zeugmann (Kyushu University) assisted him as co-chair. The Proceedings of ALT'98 have been published as ``Algorithmic Learning Theory, 9th International Conference, ALT'98, Otzenhausen, Germany, October 1998'' within the series Lecture Notes in Artificial Intelligence number 1501, Springer-Verlag, Heidelberg, ISBN 3-540-65013-X, edited by Michael M. Richter, Carl H. Smith, Rolf Wiehagen and Thomas Zeugmann. There were 45 participants from Australia, Austria, Canada, France, Germany, Italy, Japan, Latvia, Russia, Singapore, Sweden, United Kingdom and the United States of America.
At the conference banquet, John Case explained the concept of an impeachment inference machine (IIM) and its problems to converge sufficiently rapidly. In the meantime, the results of the latest elections suggest that the IIM will converge to ``no''. Charming Makoto Zeugmann visited as the youngest learning theorist this conference and enjoyed watching the billiard session after the banquet.
Steffen Lange and Thomas Zeugmann organized the Workshop on Applied Learning Theory which preceded the conference on October 7, 1998, at the University of Kaiserslautern. Masayuki Numao (Tokyo Institute of Technology, Japan) presented the invited talk on ``Learning of human feelings by inductive logic programming'': The main focus of this work was on to build a system, which selects for a Karaoke system the best pictures corresponding emotionally to the song. Then both, the songs and the pictures, where classified by humans with respect to the feelings caused by them and the learning algorithm had to find a matching for both. A further goal of the work is to choose the chords to a given melody. Further topics of the workshop were the synthesis of recursive functions from independent parameters, PAC-learning, grammatical inference in natural speech recognition, knowledge discovery in biosequences and the synthesis of intelligent pads.
Finally, let us conclude this report by announcing that the next ALT will be held together with the International Conference on Discovery Science (DS'99) in Tokyo, Japan. The conference chair is Professor Takashi Yokomori, Professor Watanabe is the chair of the program committee and Professor Taisuke Sato the local arrangements chair. More information can be found on the ALT'99 homepage.
Frank Stephan, firstname.lastname@example.org
Mathematisches Institut, Universität Heidelberg
Werner Stein, email@example.com
Arbeitsgruppe Algorithmisches Lernen, Universität Kaiserslautern.
The committee likes to thank the referees Dana Angluin, Hiroki Arimura, Jose L. Balcazar, Carlos Domingo, Sally Goldman, Tibor Hegedüs, Koichi Hirata, Eiju Hirowatari, Hiroki Ishizaka, Roni Khardon, Jyrki Kivinen, Satoshi Kobayashi, Takeshi Koshiba, M.R.K. Krishna Rao, Phil Long, Zdravko Markov, Tetsuhiro Miyahara, Yasuhito Mukouchi, Masayuki Numao, Yoshiaki Ohkubo, Seishi Okamoto, Vijay Raghavan, Ayumi Shinohara, Frank Stephan, Noriko Sugimoto, György Turan, Dawn Wilkins, Akihiro Yamamoto and Takashi Yokomori for cooperating in the reviewing process.
to the ALT'98 Proceedings Page