Dimensionality Reduction Algorithm Based on Neighborhood Rival Linear Embedding
LI Yan-ya · 2015
In order to improve the correctness of locally linear embedding caused by sparse data,a novel dimensionality reduction algorithm based on neighborhood rival linear embedding was proposed in this paper.According to the statistical information,it determines local linear dynamic range,adopts the cam distribution to find neighbors of data points,and avoids the lack of the direction of neighbor selection.In the case of sparse data sets,the algorithm can effectively obtain local and global information of data.The experiment to test the improved algorithm obtains a good effort of reducing dimension.The experimental results on the image retrieval using the Corel database show the efficiency of the algorithm.