基于哈希表的动态向量降维方法的研究及应用

摘要

提出并实现了一种简洁的基于哈希表的动态向量降维方法.该方法用哈希表作为文档特征向量的存储数据结构,省去了预先构建向量模板的环节,实现了高维次稀疏特征向量的动态降维,有效减少了分类算法的数据计算量,能够显著提高分类器的性能.%This paper proposes an algorithm based on hash table for dynamic vector dimension-reduction. The proposed algorithm uses hash table as data structure of characteristic vector so as to reduce the step of constructing vector template. This can realize dynamic vector reduced-order of high dimensional times sparse vector, and effectively reduce the computation of data classification algorithm so as to improve the capability of classifier.