Self-supervised sparse coding scheme for image classification based on low rank representation

Abstract

Recently, sparse representation, which relies on the underlying assumption that samples can be sparsely represented by their labeled neighbors, has been applied with great success to image classification problems. Through sparse representation-based classification (SRC), the label can be assigned with minimum residual between the sample and its synthetic version with class-specific coding, which means that the coding scheme is the most significant factor for classification accuracy. However, conventional SRC-based coding schemes ignore dependency among the samples, which leads to an undesired result that similar samples may be coded into different categories due to quantization sensitivity. To address this problem, in this paper, a novel approach based on self-supervised sparse representation is proposed for image classification. In the proposed approach, the manifold structure of samples is firstly exploited with low rank representation. Next, the low-rank representation matrix is used to characterize the similarity of samples in order to establish a self-supervised sparse coding model, which aims to preserve the local structure of codings for similar samples. Finally, a numerical algorithm utilizing the alternating direction method of multipliers (ADMM) is developed to obtain the approximate solution. Experiments on several publicly available datasets validate the effectiveness and efficiency of our proposed approach compared with existing state-of-the-art methods.

Tensor Analysis-based Nonlocal Magnetic Resonance Image Reconstruction with Compressed Sensing

“…In addition, [11] pointed out that the full variational regularization term TV • can make the reconstructed image more of a sparse representation and introduce this term into the CS reconstruction model [12]. For a specific 12 ,…”

Joint Cross-view Heterogeneous Discriminative Subspace Learning via Low-rank Representation

“…These methods usually project different views data into the same common subspace [8]- [10], and we find that cross-view data presents two different but intertwined structures in the original high-dimensional space, for example, a sample from cross-view data has both a class structure representing its own semantic information and a view structure containing its view information. The heterogeneous information brought by the view structure can affect the performance of retrieval and recognition.…”

Robust Graph Learning From Noisy Data

“…Recently, learning graphs from data automatically has drawn significant attention [36][37][38][39][40][41][42][43]. By this means, the most informative neighbors for each data point are automatically selected and it is free of similarity measure metrics, which are often data-dependent and sensitive to noise and outliers [44].…”