AAAI-94 SPRING SYMPOSIUM ON GOAL-DRIVEN LEARNING Stanford University March 21-23, 1994 CALL FOR PARTICIPATION Program Committee: Marie desJardins (co-chair) Lawrence Hunter Foster John Provost Ashwin Ram (co-chair) Goal-driven learning refers to the process of using the overall goals of an intelligent system to make decisions about when learning should occur, what should be learned, and which learning strategies are appropriate in a given context. This focusing process may take place at any decision point during learning---for example, when determining what to learn, selecting a bias, pruning the space of theories to be considered, or generating experiments for data gathering. Research in psychology, education, and AI has shown the need for intelligent systems to make decisions about what and how to learn. The common rationale, and the principle around which the symposium will be organized, is that the value of learning depends on how well it satisfies the goals of the system. The symposium will bring together researchers from diverse research areas to discuss issues in how learning goals arise, how they affect learner decisions of when and what to learn, and how they guide the learning process. Topics addressed by the symposium will span the diverse work in this area, which includes research in formulating learning goals, experiment generation, utility of knowledge assessment, evaluating and selecting learning biases, explanation-based learning, learning from texts, active learning, case-based reasoning, formal analyses of decision making, automated question generation, knowledge acquisition planning, reinforcement learning, and control theory. We encourage researchers from fields other than these to submit papers on related research. In addition to technical presentations, the symposium will include a session of invited talks on relevant topics, such as formal utility analyses of learning, automated experiment planning, psychological evidence for goal-driven learning behavior in humans, human educational motivation, and question generation. Depending on the number and quality of submitted papers, we may include a poster session as well. If there is sufficient overlap, we will arrange a joint session with the symposium on Decision-Theoretic Planning. Time will be set aside for debate and discussion during the technical sessions, and the symposium will conclude with a panel and audience discussion of the issues raised during the symposium. Members of the concluding panel will be selected during the meeting, with the intention of creating a panel representative of the viewpoints expressed at the symposium. In order to stimulate debate and discussion on future directions for research as well as evaluation of existing approaches to goal-driven learning, we encourage the submission of extended abstracts describing work in progress, position papers, and papers describing innovative unexplored approaches, as well as papers describing mature research results. If you wish to present a paper, please submit four hardcopies of a paper or extended abstract by October 15 to: Prof. Ashwin Ram College of Computing Georgia Institute of Technology Atlanta, GA 30332-0280 (404) 853-9372 Shorter submissions (under five pages) are encouraged; however, longer submissions (up to ten pages) will be accepted. If you wish only to participate in the workshop, please submit two hardcopies of a research summary describing your relevant research interests. Questions may directed to Ashwin Ram (ashwin@cc.gatech.edu) or to Marie desJardins (marie@erg.sri.com). SCHEDULE: October 15, 1993 Papers due November 15, 1993 Acceptance/rejection notices mailed January 31, 1994 Camera-ready papers due February 15, 1994 Registration deadline for invitees March 1, 1994 Final registration deadline March 21-23 Symposium at Stanford Univ.