Presentation | 2012-06-19 Improving the Variable Setting of Softmax Selection From an Information-Theoretic Viewpoint Kazunori IWATA, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | We focus on softmax selection which is the most popular description of the policy for action selection in reinforcement learning. Compared with other sophisticated methods in the literature, it is easy to implement and simple because there is essentially only one parameter that needs to be tuned. Moreover, it is often adequate in practice when the parameter is set appropriately for the environment. In this paper, we improve its variable setting to extend the bandwidth around the best parameter so that we can save time and cost in the implementation and parameter-tuning. Using various types of tasks, we show that our setting is effective in extending the bandwidth. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | reinforcement learning / softmax selection / information theory |
Paper # | IBISML2012-4 |
Date of Issue |
Conference Information | |
Committee | IBISML |
---|---|
Conference Date | 2012/6/12(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | |
Vice Chair | |
Secretary | |
Assistant |
Paper Information | |
Registration To | Information-Based Induction Sciences and Machine Learning (IBISML) |
---|---|
Language | ENG |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Improving the Variable Setting of Softmax Selection From an Information-Theoretic Viewpoint |
Sub Title (in English) | |
Keyword(1) | reinforcement learning |
Keyword(2) | softmax selection |
Keyword(3) | information theory |
1st Author's Name | Kazunori IWATA |
1st Author's Affiliation | Graduate School of Information Sciences, Hiroshima City University() |
Date | 2012-06-19 |
Paper # | IBISML2012-4 |
Volume (vol) | vol.112 |
Number (no) | 83 |
Page | pp.pp.- |
#Pages | 8 |
Date of Issue |