一种改进的无参数自组织映射算法

An improved parameterless self-organization map algorithm

通过允许映射对没有被较好映射的输入作较大的调整,无参数自组织映射(PLSOM)能够快速而正确地适应新的输入范围,但是输入分布与权密度之间对应性较差.论文提出了一种基于PLSOM的改进算法.在两种不同的情况下采用两种不同的权值更新方法.一种采用修改过的PLSOM,另一种则采用改进过的SOM.实验结果表明,这种改进算法不仅能快速正确地适应新的输入范围,而且能较好地体现输入分布.

By allowing the map to make large adjustments in response to inputs that were not well mapped, the parameterless self-organization map （PLSOM） could adapt correctly to the new input range quickly, but that came at the cost of lower correspondence between the input distribution and the weight density. In the paper, an imoroved algorithm was proposed based on the PLSOM. Two different weight updating methods were used in two different cases. The modified PLSOM was used in one case, and the improved SOM in another case. The experimental results showed that the imoroved algorithm could not adapt correctly to the new input range quickly, but also reflected the input distribution quite well.