Reinforcement Learning with Internal Reward for Multi-Agent Cooperation: A Theoretical Approach
Journal Title: EAI Endorsed Transactions on Collaborative Computing - Year 2016, Vol 2, Issue 8
Abstract
This paper focuses on a multi-agent cooperation which is generally difficult to be achieved without sufficient information of other agents, and proposes the reinforcement learning method that introduces an internal reward for a multi-agent cooperation without sufficient information. To guarantee to achieve such a cooperation, this paper theoretically derives the condition of selecting appropriate actions by changing internal rewards given to the agents, and extends the reinforcement learning methods (Q-learning and Profit Sharing) to enable the agents to acquire the appropriate Q-values updated according to the derived condition. Concretely, the internal rewards change when the agents can only find better solution than the current one. The intensive simulations on the maze problems as one of testbeds have revealed the following implications:(1) our proposed method successfully enables the agents to select their own appropriate cooperating actions which contribute to acquiring the minimum steps towards to their goals, while the conventional methods (i.e., Q-learning and Profit Sharing) cannot always acquire the minimum steps; and (2) the proposed method based on Profit Sharing provides the same good performance as the proposed method based on Q-learning.
Authors and Affiliations
Fumito Uwano, Naoki Tatebe, Masaya Nakata, Keiki Takadama, Tim Kovacs
Keywords
Related Articles
- EP ID EP45716
- DOI http://dx.doi.org/10.4108/eai.3-12-2015.2262878
- Views 315
- Downloads 0
Design of Pet Robots with Limitations of Lives and Inherited Characteristics
In this paper, we propose a framework of life duration and inheritance for pet robots to make them have original characteristics in their limited lives. The purpose of our research is to develop a pet robot that enables...
Message from the Editors-in-Chief
We are very pleased to welcome you all to the second issue of this journal. In this issue, we present six selected invited papers that span various aspects of collaborative computing and technologies. In particular, t...
The Richness of Open-ended Play - Rules, feedback and adaptation mechanisms in intelligent play environments
How can we design intelligent play environments for open-ended play that support richness in play? Rich play can be described as ongoing play that changes over time in character, form and nature. This paper elaborates on...
Welcome Message from the Editor-in-Chief.
No Abstract
Reconciling Schema Matching Networks Through Crowdsourcing
for data integration purposes. Although several automatic schema matching tools have been developed, their results are often incomplete or erroneous. To obtain a correct set of correspondences, usually human effort is re...