Name
Abstract | ||
|---|---|---|
Parallel operation of learning automata (LA), which is proposed by Thathachar and Arvind, is a promising mechanism that can reduce the computational burden without compromising accuracy. However, as far as we know, this parallel mechanism has not been widely used due to two reasons: one is the fact that the environment can response to multi-actions simultaneously are few, the other is the relatively slow speed of the learning process.In this paper, a novel parallel framework is presented to reduce the number of required interactions between the incorporated pursuit LA and the environment by introducing decentralized learning and centralized fusion. The philosophy is to learn various aspects of the problem at hand by taking advantage of the diverse exploration of decentralized learning and summarize the common knowledge learned by centralized fusion. Simulations are conducted to verify the effectiveness of our framework and demonstrate its outperforming. The proposed framework is further applied to the stochastic point location problem and obtains an attractive performance. |
Year | DOI | Venue |
|---|---|---|
2017 | 10.1016/j.neucom.2016.09.082 | Neurocomputing |
Keywords | Field | DocType |
Learning automata,Parallel framework,Decentralized learning,Centralized fusion,Learning speed | Learning automata,Point location,Computer science,Common knowledge,Artificial intelligence,Imagination,Machine learning | Journal |
Volume | Issue | ISSN |
228 | C | 0925-2312 |
Citations | PageRank | References |
0 | 0.34 | 22 |
Authors | ||
5 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Hao Ge | 1 | 18 | 3.65 |
| Jianhua Li | 2 | 20 | 3.19 |
| Li, S. | 3 | 16 | 2.84 |
| W.-G. Jiang | 4 | 26 | 3.43 |
| Yifan Wang | 5 | 12 | 1.87 |