DLDL: Dynamic Label Dictionary Learning via Hypergraph Regularization
This work addresses a bottleneck in dictionary learning for classification, offering a method to enhance semi-supervised and unsupervised tasks, though it appears incremental as it builds on existing label-based approaches.
The paper tackles the limited effectiveness of label-based dictionary learning in semi-supervised and unsupervised settings by proposing DLDL, which generates soft labels for unlabeled data using hypergraph regularization, achieving improved performance on remote sensing datasets.
For classification tasks, dictionary learning based methods have attracted lots of attention in recent years. One popular way to achieve this purpose is to introduce label information to generate a discriminative dictionary to represent samples. However, compared with traditional dictionary learning, this category of methods only achieves significant improvements in supervised learning, and has little positive influence on semi-supervised or unsupervised learning. To tackle this issue, we propose a Dynamic Label Dictionary Learning (DLDL) algorithm to generate the soft label matrix for unlabeled data. Specifically, we employ hypergraph manifold regularization to keep the relations among original data, transformed data, and soft labels consistent. We demonstrate the efficiency of the proposed DLDL approach on two remote sensing datasets.