摘要
讨论了基于样本最大分类信息的改进划分熵的若干性质 ,并对Bezdek提出的划分熵、基于样本最大分类信息的改进划分熵、划分熵的归一化形式的分类性能进行了比较实验。结果表明归一化形式的分类性能是最好的。
Some properties of modified partition entropy, which based on pattern's maximum classification information are discussed. Compare experiments about Bezdek's partition entropy, modified partition entropy which based on pattern's maximum classification information and normalized partition entropy are given. The results show that the classification performance of normalized partition entropy is optimal among three formulas.
出处
《西安邮电学院学报》
2002年第3期64-68,共5页
Journal of Xi'an Institute of Posts and Telecommunications
基金
国家自然科学基金资助项目 (批准号 :6 9972 0 41)
