International Conference on
Machine Learning and Data Mining MLDM´2003
Report about MLDM´2003
Renaissance Hotel Leipzig/Germany
July 5-7, 2003
Francesco Tortorella, Dipartimento di Automazione, Elettromagnetismo, Ingegneria dell'Informazione e Matematica Industriale Università degli Studi di Cassino, Italy
The 3rd International Conference on Machine Learning and Data Mining was held in Leipzig, the old and fascinating Saxon city in the centre of Germany where J.S. Bach spent half of his life and the first book was printed. The conference was sponsored by IAPR and IAPR TC17 on Machine Learning and Data Mining (http://www.ibai-research.de) and was co-chaired by Petra Perner from the Institute of Computer Vision and Applied Computer Science (IBaI) in Leipzig, Germany and Azriel Rosenfeld from the University of Maryland, USA.
About 50 researchers from 15 different countries participated in the conference. A total of 33 high quality papers were presented, selected by the program committee from 75 submissions. The topics addressed can be grouped into nine areas: support vector machines, pattern discovery, decision trees, clustering, classification and retrieval, case-based reasoning, Bayesian models and methods, association rules, applications.
The conference included two keynote invited talks. The first one was given by Susan Craw from the Robert Gordon University in Aberdeen (Scotland, UK) and focused on case-based reasoning systems. In particular, Susan described an approach to capture directly from the cases stored in the system some knowledge about how to retrieve relevant cases and how to adapt them to solve new problems. The second invited talk was given by Horst Bunke from the University of Bern (Switzerland). Horst provided a survey on algorithms devoted to graph matching and clustering. Their usefulness for object representation and recognition is well established in disciplines such as structural pattern recognition. However, the data mining field also can take great advantage of such graph-based tools, specially when the objects to be mined exhibit a spatial structure which cannot be adequately represented by a linear feature vector.
In my opinion, the most valuable quality of the conference was its interdisciplinary characteristic. The quality of the presentations and their variety made MLDM 2003 a favourable opportunity for the communities of pattern recognition, machine learning and data mining to meet together and discover common interest points: researchers coming from these different fields found an optimal place to share their ideas and find out new possible research topics. This was made possible also thanks to the careful arrangement of the sessions which gave large time to informal conversations.
The excellent technical program was completed by a very pleasant social program. In particular, participants and accompanying persons had the opportunity of attending one of the famous "Sommerkonzerte" (summer concerts) performed in the St. Thomas Church, where J.S. Bach served as "Kantor" (i.e. church music director) for more than ten years. The wonderful acoustics of the church, the outstanding performance given by the musicians together with the particular atmosphere (the tomb of J.S. Bach is placed in the chorus of the church) made that concert an unforgettable experience. The conference banquet took place in the "zum Arabischen Coffe Baum", one of the oldest restaurants in Leipzig, where we had the chance not only to taste saxonian dishes but also to visit the "Coffee Museum" which demonstrates the saxonian passion for this drink.
A final note regards the proceedings of the conference which have been published by Springer in the series "Lecture Notes in Artificial Intelligence" (LNCS/LNAI 2734, ISBN 3-540-40504-6).