Convergence and asymptotic normality of variational Bayesian approximations for exponential family models with missing values

摘要

This year marks the 20th anniversary of the Conference of Uncertainty in Artificial Intelligence (UAI). From its beginnings as a small workshop, UAI has grown to become the leading conference in the field. It is now the primary international forum for presenting new results on the use of principled methods for reasoning under uncertainty within intelligent systems. The scope of UAI is wide, including, but not limited to, representation, automated reasoning, learning, decision making, and knowledge acquisition under uncertainty. This year's conference (UAI 2004) continues the tradition, including contributions that report on advances in these core areas, as well as insights derived from the construction and use of applications involving uncertain reasoning. This volume comprises the papers accepted for presentation at UAI 2004, held at the Banff Park Inn in Banff, Canada, from July 7 through 11, 2004. Papers appearing in this volume were subjected to rigorous review; three Program Committeemembers (or in some cases, auxiliary reviewers) reviewed each paper under the supervision of an Area Chair, who made recommendations to the Program Chairs. The assignment of Program Committee members to papers was based on their expertise and expressed interests in the papers, with an eye toward coverage of the relevant aspects of each paper. This year a record 253 papers were submitted to UAI, and 76 papers were accepted for plenary or poster presentation at the conference. All accepted papers appear in this volume. We are confident that the proceedings, like past UAI Conference Proceedings, will become an important archival reference for the field.