2016 Fiscal Year Annual Research Report
Fast Optimal Transport and Applications to Inference and Simulation in Large Scale Statistical Machine Learning
| Project/Area Number |
26700002
|
|---|
| Research Institution | Kyoto University |
|---|
Principal Investigator |
Cuturi Marco 京都大学, 情報学研究科, 准教授 (80597344)
|
|---|
| Project Period (FY) |
2014-04-01 – 2017-03-31
|
| Keywords | 機械学習 / 最適輸送理論 |
|---|
| Outline of Annual Research Achievements |
The main motivation behind my work was to design new machine learning methodologies built upon optimal transport theory. The main obstacle to this goal was computational, since optimal transport is notoriously costly to compute. To avoid that issue, I proposed 3 years ago a very efficient numerical scheme to solve optimal transport problems that can scale up to large scales and use recent progresses in hardware, namely GPGPUs. This breakthrough has inspired several works in the span of 3 years, in machine learning and beyond. Many of these breakthroughs were the direct fruit of this Wakate A funding. This funding resulted in publications published in the best research forums: 2 NIPS papers, 2 SIGGRAPH papers, 2 ICML papers, 2 SIAM journal papers, the organization of a NIPS workshop, several visits and collaborations.
|
| Research Progress Status |
28年度が最終年度であるため、記入しない。
|
| Strategy for Future Research Activity |
28年度が最終年度であるため、記入しない。
|
| Causes of Carryover |
28年度が最終年度であるため、記入しない。
|
| Expenditure Plan for Carryover Budget |
28年度が最終年度であるため、記入しない。
|
