Theoretical analysis and information engineering application of approximate inference methods

点击翻译按钮获取中文摘要

Stochastic inference is important in many problems in the fields such as information engineering and statistics. Recently, more attention is paid for approximate inference methods because exact inference becomes intractable for large size problems. In this research, we have built mathematical framework to analyze some of the approximate inference methods. Using the theoretical results, we have improved the mobile reception of digital satellite broadcasting. Through the interactions within this project, we have also studied the neuron communication channel capacity, the motor planning problem in the brain, and the multi-class classification problem.