一种改进的随机近邻算法

Improved method of random adjacent classification

提出一种随机近邻分类的改进算法,它仍采用随机近邻判决准则,利用全局寻优思想,设计了最邻近生成算法,在理论和实践上说明了新算法分类良好的非线性、非球形性的特点,获得了与样本输入顺序无关的确定性结果。该算法原理简单,计算步骤明确,易于编程,可操作性强,便于应用推广。

This paper presents an improved method of random adjacent classification. It designs the most born algorithm in the adjacent still adopting the rule of random adjacent judgment and making use of the overall situation look for the excellent thought. That shows the classification of the new algorithm has a very good character of nolinear and nonspherical the oretically and practically and achieved deterministic (results) which have nothing to do with the sequence of samples reading. In the algorithm, the computing the step is clear and definite and easy to plait distance, and the manipulativity is strong and easy to apply and generaliza.