Theories of structured estimation methods for large scale data and their applications

2017 Fiscal Year Final Research Report

• Project/Area Number: 25730013
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Statistical science
• Research Institution: The University of Tokyo (2017); Tokyo Institute of Technology (2013-2016)
• Principal Investigator: SUZUKI Taiji 東京大学, 大学院情報理工学系研究科, 准教授 (60551372)
• Project Period (FY): 2013-04-01 – 2018-03-31
• Keywords: 構造的正則化 / テンソルモデリング / ベイズ推定 / 再生核ヒルベルト空間 / ガウシアンプロセス / 確率的最適化 / 高次元統計 / 統計的学習理論

Outline of Final Research Achievements

Recently, the size of dataset is getting larger and larger in several areas. Moreover, such data often contains various structures. To deal with such a complicated and large data, we have focused on structured sparsity and developed new estimation methods and computational methods in a comprehensive manner. Specifically, we have proposed a new stochastic optimization method called stochastic alternating direction method of multipliers that work efficiently for structured regularization methods. We also studied so called tensor modeling and proposed some estimators that satisfy mini-max optimality. Through the above mentioned problems, we have studied theories and applications in a comprehensive manner.

Free Research Field

機械学習