半监督增强线性近邻传递算法

SEMI-SUPERVISED INCREMENTAL LINEAR NEIGHBORHOODS PROPAGATION AGORITHM

本文将先验鉴别信息引入到降维过程中,融合线性近邻传递模型,提出了半监督增强线性近邻传递算法S-ILNP(Semi-supervised Incremental Linear Neighborhoods Propagation)。该方法首先利用先验标签信息构建类间和类内图,再依据拉普拉斯映射原理实现维数约减,运用线性近邻传递实现半监督学习,标签信息由全局一致性假设,通过局部最近临,从有标签数据点进行全局传递标注。该算法充分利用先验鉴别信息,显著提高了图像检索的准确度。

In this paper, priori information is put into the processes of dimensionality reduction, fusing the model of linear neighborhoods propagation,we propose a new semi-supervised incremental linear neighborhoods propagation algorithm. First of all, the priori label information is used to construct within-class graph and between-class graph. Secondly, Laplace eigenmaps principle is applied to achieve the goal of dimensionality reduction and then to carry out semi-supervised learning with linear neighborhoods propagation. At last, all the unlabeled points are signed the suitable labels from the labeled points by using the local linear neighborhoods with sufficient smoothness. the accuracy of image retrieval with our proposed algorithm are greatly improve by making use of the priori identification information.