摘要
本文提出了一个基于小生境遗传算法和模式缓存的时间序列中频繁结构模式的发现算法,该算法具有轻便、灵活、可扩放性好的特点,可根据实际情况合理配置计算时间和所占用的内存资源,并可实现挖掘结果的实时动态更新输出,在实际时间序列数据上的实验证明了该算法的有效性。
An Algorithm based on multiple-niche genetic algorithm and the technique of pattern caching for discovering frequent-emerging configuration patterns in time series is proposed. Using this algorithm, memory resources can be appropriately allocated, computational time can be shorten, the result of data-mining process can be updated in real time. The effectiveness of proposed algorithm is verified by the experiment on the real data of time series as to show its merits as convenient, flexible and scalable.
出处
《电路与系统学报》
CSCD
2004年第4期81-85,133,共6页
Journal of Circuits and Systems
基金
国家自然科学基金资助项目(60171029)
中国科技大学留学回国人员科研启动基金资助项目(KB2509)
关键词
时间序列
数据挖掘
频繁结构模式
遗传算法
模式缓存
time series
data mining
frequent-emerging configuration pattern
genetic algorithm
pattern caching
