首页> 外文会议>SSST 2012 >Conservation of Information in Relative Search Performance
【24h】

Conservation of Information in Relative Search Performance

机译:在相对搜索性能中保护信息

获取原文

摘要

While conservation of information popularized by the No Free Lunch (NFL) theorem establishes that all search algorithms have the same performance on the average, this appears not to be true when performance is compared in a relative manner. Some algorithms look to perform better than others. However, this advantage is lost when averaging is over a group of related algorithms. Every advantage against one algorithm is balanced by a disadvantage against a related algorithm. From this perspective, conservation of information still applies. As a consequence, comparative transitivity does not hold. If search procedure Z beats Y which, in turn, beats X, we cannot conclude that Z beats X. Indeed, the opposite might be true.
机译:虽然保护信息的储蓄不含免费午餐(NFL)定理,但在平均水平的平均来的所有搜索算法都具有相同的性能,而当以相对方式比较性能时,这似乎是不是真的。一些算法看起来比其他算法更好。但是,当平均超过一组相关算法时,这种优势丢失。对一个算法的每个优势通过针对相关算法的缺点来平衡。从这个角度来看,信息仍然适用。因此,比较传递性不会持有。如果搜索程序Z次级,这反过来又会击败X,我们不能得出结论,Z节拍X.实际上,相反可能是真的。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号