Information theory is the science of quantifying, encoding, and extracting information. The research proposed in this career development plan consists of work designed to strengthen the theoretical foundations and empirical applicability of computational learning through the development of a rigorous, information-theoretic framework for investigating this discipline. Specifically, an information-theoretic framework, based on communication theory, is proposed for investigating the probably approximately correct (PAC) model of machine learning. This framework is shown to be robust and extensible; new learning algorithms, general measures of performance, and analysis techniques are all proposed and demonstrated.
In addition to providing a means for strengthening the theoretical foundations of computational learning, this framework also provides a mechanism for developing and rigorously analyzing new learning algorithms. New algorithms for hypothesis boosting, document classification, document filtering, and meta-search are all proposed. The proposed applications work has both theoretical and experimental components: all new algorithms are to be analyzed, implemented and tested using benchmark data.
Teaching and education are an integral part of this career development plan. Both graduate and undergraduate students are introduced to topical research, both theoretical and applied, through the projects proposed. New courses on learning and information retrieval, accessible to both graduate and undergraduate students, are also proposed.
