Paper
Moratorium Effect on Estimation Values in Simple Reinforcement Learning
- Authors:
- Katsuhiro Honda; Akira Notsu; Yuki Tezuka
- Abstract
- In this paper, we introduce low-priority cut-in (moratorium) to chain form reinforcement learning, which we proposed as Simple Reinforcement Learning for a reinforcement learning agent that has small memory. In the real world, learning is difficult because there are an infinite number of states and actions that need a large number of stored memory and learning time. To solve the problem, better estimated values are categorized as “GOOD” in the reinforcement learning process. Additionally, the alignment sequence of estimated values is changed because they are regarded as an important sequence themselves. However, the method is heavily affected by the action policy. If an agent tends to search many states, its memory overflows with low-value data. Thus, low-priority cut-in (moratorium) enhances the method in order to solve this problem. We conducted some simulations and observed the influence of our methods. Several simulation results show good influence on learning.
- Keywords
- Reinforcement Learning; Q-learning; State-action Set Categorization
- StartPage
- 112
- EndPage
- 119
- Doi
- 10.5963/IJCSAI0303004