A study on the application of a source coding algorithm to statistical learning problems

2015 Fiscal Year Final Research Report

• Project/Area Number: 25730023
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Statistical science
• Research Institution: Takasaki City University of Economics (2014-2015); Waseda University (2013)
• Principal Investigator: Ishida Takashi 高崎経済大学, 経済学部, 准教授 (70409639)
• Project Period (FY): 2013-04-01 – 2016-03-31
• Keywords: 情報源符号化 / 統計的学習 / 機械学習

Outline of Final Research Achievements

In this study, we analyzed a property of entropy rate of a source models in source coding problems. An effect of model parameters to the behaviors of the value of entropy rate and the upper / lower bound of it were verified by some numerical experiments for the extended class of source model. In Addition, we analyzed an effectiveness of the prediction algorithm to which a source coding method is applied. The prediction precision of the algorithm applying to a certain class of a decision tree model was verified by some numerical experiments. Furthermore, the effectiveness of the algorithm to the actual data was also verified.

Free Research Field

統計的学習理論