Studies of models and algorithms in machine learning via submodular optimization

2017 Fiscal Year Final Research Report

• Project/Area Number: 16H06676
• Research Category: Grant-in-Aid for Research Activity Start-up
• Allocation Type: Single-year Grants
• Research Field: Mathematical informatics
• Research Institution: The University of Tokyo
• Principal Investigator: Soma Tasuku 東京大学, 大学院情報理工学系研究科, 助教 (90784827)
• Project Period (FY): 2016-08-26 – 2018-03-31
• Keywords: 組合せ最適化 / 機械学習

Outline of Final Research Achievements

Recently, submodular optimization -- a branch of combinatorial optimization -- has attracted interests in the machine learning community. In this project, we studied submodular optimization and its applications to machine learning, and obtained the following results:
1. We developed new approximation guarantee based on the concept of discrete convexity, which improves previous approaches based on the concept of curvature.
2. For dictionary learning (a problem studied in compressed sensing and machine learning), we devise a new combinatorial algorithm.

Free Research Field

組合せ最適化