Affinity Matrix Learning Via Nonnegative Matrix Factorization for Hyperspectral Imagery Clustering

Qin, Yao; Li, Biao; Ni, Weijian; Quan, Sinong; Bu, Hui

doi:10.1109/jstars.2020.3040218

Cited by 14 publications

(9 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, these methods require O(N 2 ) memory to store the affinity matrix and may suffer from an "out of memory" error in the training phase. Following [7], [12], [13], [17], [19], [23], [50], a subset of these datasets is used in our method for computational efficiency. Particularly, the subset taken from Salinas dataset is also known as Salinas-A dataset.…”

Section: Resultsmentioning

confidence: 99%

Self-Supervised Deep Subspace Clustering for Hyperspectral Images With Adaptive Self-Expressive Coefficient Matrix Initialization

Qin

Ling

et al. 2021

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

Self Cite

View full text Add to dashboard Cite

Deep subspace clustering network has shown its effectiveness in hyperspectral image (HSI) clustering. However, there are two major challenges that need to be addressed: 1) lack of effective supervision for feature learning; 2) negative effect caused by the high redundancy of the global dictionary atoms. In this paper, we propose an end-to-end trainable network for HSI clustering. Specifically, to ensure the extracted features are wellsuited to subsequent subspace clustering, the cluster assignments with high confidence are employed as pseudo-labels to supervise the feature learning process. Then, an adaptive self-expressive coefficient matrix initialization strategy is designed to reduce the dictionary redundancy, where the spectral similarity between each target sample and its neighbors are modeled via the knearest neighbor graph to guide the initialization. Experimental results on three public HSI datasets demonstrate the effectiveness of the proposed method. In particular, our method outperforms several state-of-the-art HSI clustering methods, and achieves overall accuracy (OA) of 100% on both SalinasA and Pavia University datasets.

show abstract

Section: Resultsmentioning

confidence: 99%

Self-Supervised Deep Subspace Clustering for Hyperspectral Images With Adaptive Self-Expressive Coefficient Matrix Initialization

Qin

Ling

et al. 2021

IEEE J. Sel. Top. Appl. Earth Observations Remote Sensing

Self Cite

View full text Add to dashboard Cite

show abstract

“…Thus, metrics, or UFE methods, achieving SOTA performance of computing affinity between HSI samples are considered to be candidates for the generation of N2P. Among these methods, local covariance matrix representation, exploiting spectral-spatial information, has been successfully applied to both unsupervised and supervised HSIC [24,25,69]. Compared with SAM and Euclidean distances, it exploited the similarities and variances of local samples and proved to be suitable for measuring distances between HSI samples.…”

Section: Generatation Of N2pmentioning

confidence: 99%

“…Since HSI annotation usually requires extensive field data collection campaigns, that are costly and impractical when the HSI scene involves incooperative areas [24], a few published works focus on unsupervised classification, i.e., clustering, that directly models the intrinsic characteristic of HSI samples to form several clusters [25]. Typical HSI clustering methods include k-means, fuzzy c-means and etc.…”

Section: Introductionmentioning

confidence: 99%

Nearest Neighboring Self-Supervised Learning for Hyperspectral Image Classification

Qin

Zhao

et al. 2023

Remote Sensing

Self Cite

View full text Add to dashboard Cite

Recently, state-of-the-art classification performance of natural images has been obtained by self-supervised learning (S2L) as it can generate latent features through learning between different views of the same images. However, the latent semantic information of similar images has hardly been exploited by these S2L-based methods. Consequently, to explore the potential of S2L between similar samples in hyperspectral image classification (HSIC), we propose the nearest neighboring self-supervised learning (N2SSL) method, by interacting between different augmentations of reliable nearest neighboring pairs (RN2Ps) of HSI samples in the framework of bootstrap your own latent (BYOL). Specifically, there are four main steps: pretraining of spectral spatial residual network (SSRN)-based BYOL, generation of nearest neighboring pairs (N2Ps), training of BYOL based on RN2P, final classification. Experimental results of three benchmark HSIs validated that S2L on similar samples can facilitate subsequent classification. Moreover, we found that BYOL trained on an un-related HSI can be fine-tuned for classification of other HSIs with less computational cost and higher accuracy than training from scratch. Beyond the methodology, we present a comprehensive review of HSI-related data augmentation (DA), which is meaningful to future research of S2L on HSIs.

show abstract

“…However, Spectral GCN Graph Convolutional Networks based Non-Negative Matrix Factorization aware Comm has a high time complexity due to the need to calculate the eigenvalues of the graph Laplacian matrix. To improve the computational efficiency of Spectral GCN, literature [34] proposed the Chebyshev network ChebyNet. ChebyNet uses Chebyshev expansion to approximate convolution operation, which significantly improves the efficiency of the graph convolution operation.…”

Section: Graph Convolutional Networkmentioning

confidence: 99%

Graph Convolutional Networks based Non-Negative Matrix Factorization aware Community detection

Scholar

Professor

2023

Preprint

View full text Add to dashboard Cite

Nonnegative matrix factorization (NMF) is widely used in community discovery because of its effectiveness and easy interpretability. However, most of the existing NMF-based community detection methods are linear. They cannot effectively deal with the nonlinear characteristics of complex networks, resulting in further improvement in community detection performance. Aiming at this problem, a convolution graph network (GCN) enhanced nonlinear NMF community discovery method NMFGCN is proposed. NMFGCN consists of two main modules: GCN and NMF, where GCN is used to learn network node representations, and NMF uses node representations as input to obtain network community representations. In addition, a joint optimization method is proposed to train NMFGCN, which enables NMFGCN to have nonlinear feature representation capabilities and enables GCN and NMF to promote each other and obtain better community segmentation results. Many experiments on artificial synthetic networks and entire networks show that NMFGCN is superior to current NMF-based community discovery methods , thus proving that NMFGCN can improve the performance of NMF community discovery methods. In addition, NMFGCN also outperforms Deep Walk and LINE standard graph representation learning methods.

show abstract

Affinity Matrix Learning Via Nonnegative Matrix Factorization for Hyperspectral Imagery Clustering

Cited by 14 publications

References 41 publications

Self-Supervised Deep Subspace Clustering for Hyperspectral Images With Adaptive Self-Expressive Coefficient Matrix Initialization

Self-Supervised Deep Subspace Clustering for Hyperspectral Images With Adaptive Self-Expressive Coefficient Matrix Initialization

Nearest Neighboring Self-Supervised Learning for Hyperspectral Image Classification

Graph Convolutional Networks based Non-Negative Matrix Factorization aware Community detection

Contact Info

Product

Resources

About