模式识别中的Oracle复杂性和非传递性
Oracle Complexity and Nontransitivity in Pattern Recognition
摘要:
识别过程的不同数学模型是已知的。本文将模式识别算法看作是图灵机上的预言机计算。这种观点在模式识别以及递归理论中似乎是有用的。递归理论在模式识别中的应用表明了识别算法比较问题与预言计算复杂性问题之间的联系。这是因为在许多情况下,我们可以只考虑符号计算的数量,或者换句话说,所需的Oracle信息量。因此,识别算法偏好问题可以表述为一个预言计算的复杂性优化问题。此外,在一组识别算法上引入某种“自然”偏好关系,我们发现它是非传递的。这与概率论中著名的非传递性悖论有关。关键词:模式识别,递归理论,非传递性,偏好关系
英文摘要:
Different mathematical models of recognition processes are known. In the present paper we consider a pattern recognition algorithm as an oracle computation on a Turing machine. Such point of view seems to be useful in pattern recognition as well as in recursion theory. Use of recursion theory in pattern recognition shows connection between a recognition algorithm comparison problem and complexity problems of oracle computation. That is because in many cases we can take into account only the number of sign computations or in other words volume of oracle information needed. Therefore, the problem of recognition algorithm preference can be formulated as a complexity optimization problem of oracle computation. Furthermore, introducing a certain "natural" preference relation on a set of recognizing algorithms, we discover it to be nontransitive. This relates to the well known nontransitivity paradox in probability theory. Keywords: Pattern Recognition, Recursion Theory, Nontransitivity, Preference Relation
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。