DLDL: Dynamic Label Dictionary Learning via Hypergraph Regularization

TL;DR

This paper introduces DLDL, a novel dictionary learning method that uses hypergraph regularization to generate soft labels for unlabeled data, improving semi-supervised learning in classification tasks.

Contribution

The paper proposes a dynamic label dictionary learning algorithm that effectively incorporates unlabeled data through hypergraph regularization, enhancing semi-supervised classification.

Findings

Demonstrates improved classification accuracy on remote sensing datasets.

Effectively generates soft labels for unlabeled data.

Maintains data relations via hypergraph manifold regularization.

Abstract

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.