Study on improved LLE algorithm based on a sample set of well-distributed and weights matrix

Song Fei, Cui Zhe

COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(4) 15-18

Chengdu Computer Institute of Chinese Academy of Science

There are large amounts of data has accumulated along with technology of computer, information and network developed. How can we using these data and mining out the valuable information are hot topics in information processing field. There are some distress and difficulties caused by the high-dimensional data on data modelling and data analysis. In this paper, a local linear embedding algorithm based on the improved uniform sample set and the weight value matrix is proposed. The test shows that the improved dimensionality reduction algorithm accuracy is significantly higher than the original LLE algorithm.