Received: from ics.uci.edu by Paris.ics.uci.edu id aa11174; 4 Jan 93 11:47 PST Received: from ics.uci.edu by q2.ics.uci.edu id aa27802; 4 Jan 93 11:46 PST To: ML-LIST: ; Subject: Machine Learning List: Vol. 5, No. 1 Reply-to: ml@ics.uci.edu Date: Mon, 04 Jan 93 11:37:55 -0800 From: Michael Pazzani Message-ID: <9301041146.aa27802@q2.ics.uci.edu> Machine Learning List: Vol. 5, No. 1 Monday, January 4, 1993 Contents: IJCAI-93 Workshop on Machine Learning and Knowledge Acquisition Minimum Description Length & Transformations in Machine Learning The Machine Learning List is moderated. Contributions should be relevant to the scientific study of machine learning. Mail contributions to ml@ics.uci.edu. Mail requests to be added or deleted to ml-request@ics.uci.edu. Back issues may be FTP'd from ics.uci.edu in pub/ml-list/V/ or N.Z where X and N are the volume and number of the issue; ID: anonymous PASSWORD: ---------------------------------------------------------------------- Date: Tue, 29 Dec 92 13:51:08 EST From: Gheorghe Tecuci Subject: IJCAI-93 Workshop on Machine Learning and Knowledge Acquisition CALL FOR PAPERS IJCAI-93 WORKSHOP MACHINE LEARNING AND KNOWLEDGE ACQUISITION: Common Issues, Contrasting Methods, and Integrated Approaches 29 August 1993, Chambery, France Machine learning and knowledge acquisition share the common goal of acquiring and organizing the knowledge of a knowledge-based system. However, each field has a different focus, and most research is still done in isolation from each other. The focus of knowledge acquisition has been to improve and partially automate the acquisition of knowledge from human experts. In contrast, machine learning focuses on mostly autonomous algorithms for acquiring or improving the organization of knowledge, often in simple prototype domains. Also, in knowledge acquisition, the acquired knowledge is directly validated by the expert that expresses it, while in machine learning, the acquired knowledge needs an experimental validation on data sets independent of those on which learning took place. As machine learning moves to more 'real' domains, and knowledge acquisition attempts to automate more of the acquisition process, the two fields increasingly find themselves investigating common issues with complementary methods. However, lack of common research methodologies, terminology, and underlying assumptions often hinder a close collaboration. The purpose of this symposium is to bring together machine learning and knowledge acquisition researchers in order to facilitate cross-fertilization and collaboration, and to promote integrated approaches which could take advantage of the complementary nature of machine learning and knowledge acquisition. Topics of interest include, but are not limited to, the following: Case Studies Case studies of integrated ML/KA methods, with analysis of successes/failures; integrated architectures for ML and KA; interactive learning systems, automated knowledge acquisition systems; Comparative Studies Comparative studies of KA and ML methods solving similar problems (e.g., knowledge base refinement methods in KA versus theory revision methods in ML, constructive induction in ML versus knowledge elicitation in KA). Analysis of the complementarity of the KA and ML approaches to knowledge base construction (e.g. KA primarily addresses the problems of KB elicitation and refinement, while ML primarily addresses issues of KB refinement and optimization). Hard Problems Analysis of hard problems in KA or ML that could be simplified by employing techniques from the other area, as well as presentation of specific solutions (e.g. the problem of new terms in ML could be simplified by employing knowledge elicitation techniques developed in KA; the credit/blame assignment problem in ML could be simplified by employing knowledge refinement techniques developed in KA; KA of problem solving rules could be automated by using apprenticeship learning techniques); Knowledge Representation Knowledge representation issues in KA and ML (adequate representations for KA, adequate representations for ML, approaches to knowledge representation in integrated ML/KA systems like translation between representations, common representations, etc.); Key Issues Key issues in ML or KA (e.g. dynamic versus static knowledge acquisition or learning, the role of explanations in KA and ML, the validation of knowledge in KA and ML); Overviews Overviews of the state-of-the-art of ML, KA or of the integration of ML and KA, Position Papers Position papers on methodology for integrated ML/KA systems or on improving the collaboration between the ML and KA communities. It is recommended that the papers make explicit the research methodology, the underlying assumptions, definitions of technical terms, important future issues, and potential points of collaboration. They should not exceed 15 pages. The organizers intend to publish a selection of the accepted papers as a book or the special issue of a journal. They encourage the authors to take this into account while preparing their papers. The format of the workshop will be paper sessions with discussion at the end of each session, and a concluding panel on the integrated approaches, guidelines for successful collaboration, and concrete action items. The number of the participants to the workshop is limited to 40. Each workshop attendee must also register for the IJCAI conference and must pay an additional 300FF (about $60) fee for the workshop. One student attending the workshop and being in charge of taking notes will be exempted from the additional 300 FF fee. Volunteers are invited. WORKSHOP Co-CHAIRS Smadar Kedar Yves Kodratoff Gheorghe Tecuci NASA Ames & Inst.for CNRS & Universite George Mason Univ.& Learning Sciences de Paris-Sud Romanian Academy (kedar@ils.nwu.edu) (yk@lri.lri.fr) (tecuci@aic.gmu.edu) PROGRAM COMMITTEE Ray Bareiss, Institute for the Learning Sciences Catherine Baudin, NASA Ames Guy Boy, European Inst. of Cognitive Sciences and Eng. Brian Gaines, University of Calgary Matjaz Gams, Jozef Stefan Institute Jean-Gabriel Ganascia, Univ. Pierre and Marie Curie Nathalie Mathe, European Space Agency and NASA Ames Ryszard Michalski, George Mason University Raymond Mooney, University of Texas at Austin Katharina Morik, Dortmund University Mark Musen, Stanford University Michael Pazzani, Univ. of California at Irvine Luc De Raedt, Catholic University of Leuven Alan Schultz, Naval Research Laboratory Mildred Shaw, University of Calgary Maarten van Someren, University of Amsterdam Walter Van de Velde, University of Brussels ADDRESS FOR CORRESPONDENCE Gheorghe Tecuci Artificial Intelligence Center, Computer Science Department George Mason University, 4400 University Dr., Fairfax, VA 22030 email: mlka93@aic.gmu.edu, fax: (703)993-3729 SUBMISSIONS Four copies of the papers (five to fifteen pages in length) should arrive at the above address by March 31, 1993. Notification of acceptance or rejection will be sent by May 10. Final papers should arrive by June 10, 1993. Those who would like to attend without a presentation should send a one to two-page description of relevant research interests and a list of selected publications. ------------------------------ Subject: Minimum Description Length & Transformations in Machine Learning From: aboulang@bbn.COM Date: Sat, 2 Jan 93 19:00:10 EST Minimum Description Length & Transformations in Machine Learning Or, Is there a Principle of Least Action for Machine Learning? In this short note I want to posit that MDL-like methodologies will become the unifying "Least Action Principles" of machine learning. Furthermore, machine learning architectures will evolve to include a fundamental capability for doing coordinate transformations and this capability will be intimately tied to the use of MDL-like methodologies in Machine Learning. By MDL-like methodologies I mean the use information-theoretic metrics on the results of any machine learning algorithm in its generalization phase. This metric is used a a decision criterion for over training by comparing the MDL-like metric of the results or the machine learning algorithm against the data itself. MDL-like methodologies are applicable to supervised and unsupervised learning. What I want to mean by the term "MDL-like" is that there is an applicable body of work in this area -- including the work of Wallace, Akaike and Rissanen. It is possible to use MDL-like metrics in the generation phase as well. Transformations and Machine Learning Many paradigmnamic problems in machine learning become "embarrassingly" simple under straightforward coordinate transformations. For instance, the two spirals problem becomes two simple lines under a polar coordinate transformation. Much of the activity of a physicist is in examination of appropriate coordinate system hosting of the problem to exploit symmetries of the problem. I posit that at least one phase of any machine learning system should include a search for appropriate coordinate system hosting. These transformations come in many different colors. For example, temporal differences is a relativising transformation in time coordinates. Another example is the growing use of wavelets for time-frequency features. A significant contributor to the complexity of the description of a problem is its chosen coordinate-system hosting. Coordinate transformations can be of two types: local and global. An example of a global transformation is the aforementioned polar hosting for the two spirals problem. The Fukashima network makes use of local transformations for robust pattern recognition. MDL can be used as the selection criteria in the transformation search. MDL as a Least Action Principle for Machine Learning MDL-like methods holds a promise to be a unifying principle in machine learning -- much like Lagrangian methods that make use of action and its minimization is *the* unifying approach in physics, cutting across classical physics, relativistic physics, and quantum mechanics. MDL-like metrics are a type of *action* for machine learning. (In fact for certain types of search in machine learning, Lagrangian optimization can be used.) (Recent work in machine vision at MIT has suggested the use of MDL as a principle for 3-d object recognition and disambiguation. It is posited that what is perceived is related to a MDL description of the 3d-scene. By the way, who is doing this work?) There are a couple of long-standing conceptual issues in machine learning: The relationship between learning methodologies - supervised, unsupervised, reinforcement learning, etc. Somehow, one would like a unifying framework for all of them. The fact that MDL-like methods can be used in several methodologies means that it could help in building such a framework. The relationship between optimization and machine learning. MDL-like metrics are posited to be the *general* optimization criterion for machine learning. MDL has broad applicability in machine learning. It can be used to guide search in both unsupervised and supervised learning. It can be used as the common optimization criterion for "multi-algorithm machine learning systems". Finally it can be used to tie the search in feature space with that of the search for coordinate system hosting. Seeking a higher form for machine learning, Albert Boulanger aboulanger@bbn.com ------------------------------ End of ML-LIST (Digest format) ****************************************