Bayesian Ying-Yang Harmony Learning, Information Theory and Information Geometry

L Xu
DOI: https://doi.org/10.1109/ijcnn.2001.938807
2001-01-01
Abstract:Learning tasks are viewed from the fundamental relationships among the observed data. The key issues of Bayesian Ying Yang system and harmony learning theory are summarized from a systematical perspective. Further discussions are made on its relations to maximum likelihood learning, information theory, information geometry and Helmholtz machine learning.
What problem does this paper attempt to address?