2012 Fiscal Year Final Research Report
Progressive research on the exploitation-oriented learning XoL
| Project/Area Number |
22500143
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | National Institution for Academic Degrees and University Evaluation |
|---|
Principal Investigator |
MIYAZAKI Kazuteru 独立行政法人大学評価・学位授与機構, 研究開発部, 准教授 (20282866)
|
|---|
| Project Period (FY) |
2010 – 2012
|
| Keywords | 経験強化型学習 / 強化学習 / 報酬と罰の設計指針 |
|---|
| Research Abstract |
This research has completed an Exploitation-oriented Learning (XoL) method that can treat multiple rewards and penalties. Furthermore the design guideline of rewards and penalties on the XoL method has been proposed through illustrative examples, namely, a course classification task, a waist-trajectory learning task for a tendon-driven biped robot, and a Keepaway task in a multi-agent environment. It claim that XoL surpass traditional Reinforcement Learning based on Dynamic Programming in application to real-world problem.
|
