The common batch manifold learning algorithms can’t achieve dimensionality reduction rapidly of additional samples with the learned manifold structure, the incremental orthogonal neighborhood preserving embedding (IONPE) manifold learning algorithm was proposed. It achieves dynamic incremental learning for the additional samples with the block processing idea based on orthogonal neighborhood preserving embedding. Firstly, select some overlapping points from the original samples and add them to the additional samples; Secondly, get the subset of low-dimensional embedding coordinates of additional samples by ONPE not depending on the original samples; Finally, based on the principle of minimizing the differences of the overlapping point coordinates, the low-dimensional embedding coordinates of the additional samples were integrated into the original samples by rotating, shifting and scaling transformation. The fault diagnosis case of the gearbox confirmed that the IONPE algorithm has good incremental learning ability. It improves the processing efficiency of the additional samples while inheriting the superior clustering performance of ONPE.
宋 涛;汤宝平;邓蕾. 动态增殖流形学习算法在机械故障诊断中的应用[J]. , 2014, 33(23): 15-19.
Song Tao;Tang Baoping;Deng Lei. A Dynamically Incremental Manifold Learning Algorithm and Its Application in Fault Diagnosis for Machinery. , 2014, 33(23): 15-19.