Indicator-based multi-objective genetic programming for workflow scheduling problem
| Autor: | Qin-zhe Xiao, Jinghui Zhong, Zhi-Hui Zhan, Wen-Neng Chen, Jun Zhang |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
020203 distributed computing
Hardware_MEMORYSTRUCTURES Computer science business.industry Distributed computing Genetic programming 02 engineering and technology Machine learning computer.software_genre Multi-objective optimization Multi objective genetic programming Workflow 0202 electrical engineering electronic engineering information engineering Key (cryptography) Workflow scheduling 020201 artificial intelligence & image processing Artificial intelligence Heuristics business Gene expression programming computer |
| Zdroj: | GECCO (Companion) |
| DOI: | 10.1145/3067695.3075600 |
| Popis: | This paper proposes an Indicator-Based Multi-objective Gene Expression Programming (IBM-GEP) to solve Workflow Scheduling Problem (WSP). The key idea is to use Genetic Programming (GP) to learn heuristics to select resources for executing tasks. By using different problem instances for training, the IBM-GEP is capable of learning generic heuristics that are applicable for solving different WSPs. Besides, the IBM-GEP can search for multiple heuristics that have different trade-offs among multiple objectives. The IBM-GEP was tested on instances with different settings. Compared with several existing algorithms, the heuristics found by the IBM-GEP generally perform better in terms of minimizing the cost and completed time of the workflow. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|