作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2010, Vol. 36 ›› Issue (10): 215-217. doi: 10.3969/j.issn.1000-3428.2010.10.074

• 人工智能及识别技术 • 上一篇    下一篇

基于概率的自适应学习预测策略

何可佳   

  1. (宁波工程学院电子与信息工程学院,宁波 315016)
  • 出版日期:2010-05-20 发布日期:2010-05-20

Probability-based Adaptive Learning Prediction Strategy

HE Ke-jia   

  1. (School of Electronic and Information Enginerring, Ningbo University of Technology, Ningbo 315016)
  • Online:2010-05-20 Published:2010-05-20

摘要: 动态电源管理技术降低系统功耗的主要办法是根据工作负载的变化动态地切换目标设备工作模式。针对自适应学习树模型的缺陷,提出基于概率的自适应学习预测策略,通过概率描述设备行为,能够提高预测正确率,从而达到系统功耗与性能之间的优化平衡。基于概率的自适应学习预测策略是一种集预测、控制、反馈为一体的预测策略。实验结果表明,该预测策略具有较好的稳定性,与其他预测策略相比可以进一步降低系统的功耗。

关键词: 动态电源管理, 预测, 自适应学习树, 基于概率的自适应学习

Abstract: Dynamic Power Management(DPM) aims at minimization of power consumption of electronic systems by dynamically switching the power state of power manageable components according to the variations of workloads. Probability-based Adaptive Learning Tree(ALT) is provided against the defect of adaptive learning tree model. By characterizing the device activity in probability, probability-based ALT has higher hit ratio and can optimize the balance between performance and power. It is one kind of prediction strategy, with collection of prediction, control and feedback. Experimental result indicates that the probability-based ALT forecast strategy has the very good stability, comparied with other prediction strategies, it may further reduce system’s power loss.

Key words: Dynamic Power Management(DPM), prediction, Adaptive Learning Tree(ALT), probability-based adaptive learning

中图分类号: