| | 講 題 | Adaptive Workflow Scheduling Based on Machine Learning Technique | | | 講 者 | Department of Computer Science1 National Taichung University of Education Taichung, Taiwan Kuan Fu Chen | | | 日 期 | 2018/10/25 | 長 度 | 00:09:43 | 人 氣 | 94 次 | | | Workflow scheduling has long been known to be a challenging
NP-complete problem. Therefore, many heuristic-based
scheduling methods have been proposed to tackle the problem.
However, although some recent efficient methods have been
shown to outperform other previous methods in average
performance, none of them can keep achieving the best
performance than others for every workflow in the experiments.
This paper presents our experience in applying machine learning
technique to develop an adaptive workflow scheduling approach,
which can dynamically choose a most appropriate scheduling
method for a specific workflow to be scheduled, for improving
workflow execution performance further. Our goal is to develop
a new scheduling method which can outperform existing ones for
as many workflows as possible, measured by win% in the
experiments. Experimental results show that our adaptive
scheduling method can achieve a much higher win% than using
any other single scheduling method for all workflows.
| | | 提 供 | TANET台灣網際網路研討會-TANET2018 | | |
|
