摘要
对交互熵理论进行了研究,提出了对称交互熵的概念,并论证了它是一种距离测度,可以用以度量两个随机变量的差异程度,我们把它作为信息特征压缩的类别可分性判据,称之为对称交互熵判据(SCEC),建立了基于SCEC的信息特征压缩算法.模拟应用表明,提出的算法是一种有效的、可靠的算法,为模式识别理论的研究提供了一种新的数据压缩方法.
The cross entropy theory is discussed, a new concept of Symmetric Cross Entropy Criterion (SCEC) is proposed and proved that SCEC is a distance measure. SCE can be used to measure the difference degree between the random variables and regarded it as class separability criterion for information feature compression, and called Symmetric Cross Entropy Criterion (SCEC). Based on SCEC, a new algorithm of information feature compression is set up. The rersults of simulation application show that the algorithm proposed here is effective and reliable. It provides a new reasearch approach of data compression for pattern recognition.
出处
《小型微型计算机系统》
CSCD
北大核心
2005年第7期1202-1205,共4页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(40074001
60435010)资助
山东省作物生物学重点实验室开放基金资助
关键词
交互熵
SCEC
信息特征压缩
模式识别
SCE
SCEC
information feature compression
pattern recognition