数据挖掘中LLE算法的改进
Improvement of LLE Algorithm in Data Mining
局部线性嵌入 (Locally Linear Embedding, LLE)算法可以对非线性数据进行降维处理，但无法有效的处理密度不均匀的数据。本文采用改进的Dijkstra距离和K-邻居图两方面改进LLE算法邻接点的选取规则,提出DKLLE算法，提高了算法对密度不均匀数据和噪声数据降维处理的准确性。通过方针实验证明了DKLLE算法在处理密度不均匀数据时的有效性和鲁棒性，在数据挖掘中比单纯的LLE算法具有更好地适应性。
Local linear embedding(LLE) algorithm can effectively reduce the nonlinear data, but it can not effectively deal with the uneven density of data. In this paper, we proposed DKLLE, an improved algorithm from LLE using improved Dijkstra distance and K-neighbor graph algorithm.The experimental results show that DKLLE can significantly improve the efficiency and robustness of the dimension reduction when dealing with uneven density,it has better adaptability than the pure LLE algorithm in data mining.
杨杨、陈永胜、邱雪松
计算技术、计算机技术
局部线性嵌入迪杰斯特拉邻居图
locally linear embeddingdijkstraneighbor graph
杨杨,陈永胜,邱雪松.数据挖掘中LLE算法的改进[EB/OL].(2015-11-27)[2025-08-02].http://www.paper.edu.cn/releasepaper/content/201511-700.点此复制
评论