Loading...
Please wait, while we are loading the content...
Similar Documents
Ecient Implementation of Distributed Genetic Algorithms on Network of Workstations
| Content Provider | Semantic Scholar |
|---|---|
| Author | Yoshida, Norihiko Araki, Ryoji |
| Copyright Year | 2007 |
| Abstract | With the advance of networking technologies, it is now getting feasible to build a parallel computation system on workstations (or even PC's) connected by high-speed network at the much less cost than on a specially-built multi-processor machine. Such system is called Network of Workstations (NoW) or Workstation Cluster. However, communication over high-speed network is still slower than over specially-built interprocessor buses in a multi-processor machine. Therefore, in order to achieve high performance, it is crucial for parallel systems on NoW to design communication schemes carefully. The number of messages and the frequency of communication must be kept low. In multi-meme-based distributed genetic algorithm systems, inter-processor communication, i. e. inter-meme migration a ects the overall convergence property. In this paper, we present a dynamic adaptation scheme for the frequency of inter-meme migration. We built a multi-meme-based distributed GA platform on Ethernet-based NoW, and through considerations and experiments, we prove that our scheme achieves better performance than xed frequency migration schemes without affecting the overall convergence properties. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://www.cs.cis.nagasaki-u.ac.jp/~yoshida/pub/soco97b.ps.gz |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
