Long and Medium-Term Research: Information-Based Approachesto Learning Relational Concepts

该奖项是根据外国卓越中心长期和中期研究计划颁发的，该计划允许美国科学家和工程师在国外公认的卓越研究中心进行为期3至12个月的研究。该项目的奖项提供了联合研究的机会，并提供了使用国外独特或互补的设施、专业知识和实验条件的机会。该奖项将支持加州大学欧文分校的迈克尔·帕扎尼博士与悉尼大学的J·罗斯·昆兰博士合作进行为期三个月的博士后研究访问。Pazzani博士将与Quinlan博士就以下主题进行合作，这些主题涉及基于信息的关系概念学习方法：A.调查处理训练数据中的噪声。他将研究最小描述长度方法的改进，以及处理噪声训练数据的替代方法。将探索的一种替代方案是对多个模型进行平均的技术。B.将关系概念学习应用于自然产生的数据集上的更大问题，识别其他研究问题，并检查Pazzani&amp；Kibler(正在出版)中的复杂性分析如何预测学习方法将如何扩展到更大的问题。该奖励建议提供国际旅行经费和三个月的津贴。

This award is under the Long and Medium-Term Research at Foreign Centers of Excellence Program, which enables U.S. scientists and engineers to conduct three to twelve months of research abroad at research centers of proven excellence. The program's awards provide opportunities for joint research, and the use of unique or complementary facilities, expertise and experimental conditions abroad. This award will support a three-month postdoctoral research visit by Dr. Michael Pazzani of the University of California, Irvine to work with Dr. J. Ross Quinlan of the University of Sydney. Dr. Pazzani will collaborate with Dr. Quinlan on the following topics involving information-based approaches to relational concept learning: A. Investigating approaches to dealing with noise in training data. He will investigate refinements of a Minimum Description Length approach, as well as alternative approaches to dealing with noisy training data. One alternative that will be explored is the technique of averaging multiple models. B. Applying relational concept learning to larger problems on naturally occurring data sets, identifying additional research issues, and examining how well the complexity analysis in Pazzani & Kibler (in press) predicts how the learning methods will scale to larger problems. The award recommendation provides funds to cover international travel and a stipend for three months.