Pith. sign in

REVIEW 5 major objections 8 minor 50 references

Robust Subspace Discovery by Block-diagonal Adaptive Locality-constrained Representation

T0 review · 5 major / 8 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read rBDLR jointly recovers multi-subspace structures and adaptive locality-preserving salient features.

desk verdict Competent incremental combination of low-rank and block-diagonal representation; the 'discovery' claim is undercut by an oracle k and imposed block structure. read the letter →

arxiv 1908.01266 v1 pith:OMY3QK5F submitted 2019-08-04 cs.CV

Zhao Zhang , Jiahuan Ren , Sheng Li , Richang Hong , Zhengjun Zha , Meng Wang This is my paper · ORCID
classification cs.CV
keywords subspacerecoverylow-rankrepresentationblock-diagonalregularizeradaptivelocalitylatentunsupervisedlearningimagerecognitionclustering
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper proposes an unsupervised representation-learning model, rBDLR, that tries to solve two jobs at once: recovering the multi-subspace structure hidden in noisy image data and extracting salient features that preserve local geometry. The authors claim that by performing latent low-rank coding in a cleaned data space, learning an adaptive block-diagonal weight matrix from the salient features, and coercing the coefficient matrix to match that weight matrix, both jobs improve each other. If the claim holds, a single objective can replace the usual two-stage pipeline of denoising-plus-coding and feature extraction, with better recognition and clustering accuracy on face, object, and digit datasets. The paper reports consistent gains over comparison methods on UMIST, COIL100, ETH80, and USPS, including under added Gaussian noise.

What carries the argument

The central object is the k-block-diagonal regularizer applied to an adaptive weight matrix W, rather than to an arbitrary representation matrix. rBDLR defines W from the salient features PU by minimizing the reconstruction error of PU under the block-diagonal constraint, which makes W interpretable and locality-preserving, and then approximates the coefficient matrix Z by W through a flexible term. The other load-bearing pieces are the Frobenius-norm latent decomposition for efficiency, the recovered clean space U = X - E for robustness to outliers, and the L2,1-norm error correction. Together these force the coefficient matrix into a strict block-diagonal form while allowing information to propagate between salient features and codes.

What would settle it

Run rBDLR on a clustering benchmark with k deliberately set to the true class count and to values above and below it. If clustering accuracy and block purity do not drop when k is wrong, the block-diagonal regularizer is not carrying the claimed load. Alternatively, test on data where the underlying subspaces do not correspond to class labels; if the recovered block structure still aligns with classes, the method is exploiting labels implicitly.

Watch

Extended reading notes

Core claim

rBDLR claims that subspace structure and adaptive locality-preserving features can be recovered jointly by minimizing one objective. The model extends Frobenius-norm latent low-rank representation by correcting errors in the data first, factorizing the cleaned data U into principal features UZ and salient features PU, then learning a nonnegative symmetric weight matrix W from the salient features under a k-block-diagonal regularizer. A flexible penalty forcing Z close to W (up to a bias) transfers the block structure onto the coefficients Z. The paper's central discovery claim is that this joint formulation yields strict block-diagonal coefficients that characterize subspace structure more accurately than codes from LatLRR, FLLRR, or BDR, and that the extracted features generalize better for recognition and clustering.

Load-bearing premise

The load-bearing premise is that the number of blocks k equals the number of true classes, which is a label-derived input in an otherwise unsupervised method; a secondary gap the paper concedes is that the alternating optimization has no proven global convergence for more than two blocks.

Editorial extensions

If this is right

  • rBDLR can be trained once on cleaned data and applied to new samples through the learned projection, avoiding the transductive recomputation of methods like RPCA and LRR.
  • Because it uses the Frobenius norm instead of the nuclear norm, the optimization avoids SVD, so the method scales better than nuclear-norm latent low-rank models on larger datasets.
  • The block-structure constraint plus adaptive weighting makes the coefficient matrix strictly block-diagonal, which should improve subspace clustering over methods whose off-diagonal codes contain inter-class connections.
  • Under Gaussian corruption, rBDLR degrades more slowly than comparison methods, indicating that performing representation in the recovered clean space adds robustness.
  • The learned salient features preserve local geometry adaptively, improving 1NN recognition over locality-preserving projections and low-rank baselines.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A natural extension the paper does not develop: since the adaptive weight matrix W is interpretable, it could be used directly as an affinity graph for spectral clustering, making the block-diagonal structure visible to downstream methods.
  • The reliance on k suggests a testable self-tuning variant: estimate k from the eigengap of the Laplacian of W, then compare with the oracle-k setting. If performance holds without labels, the method becomes truly unsupervised.
  • The same joint recovery-and-feature-extraction design could be applied to video background modeling or multi-view data, where the block structure corresponds to scenes or views rather than subject classes.
  • The paper itself notes that the alternating optimization has no proven global convergence for more than two blocks, so a monotone-decreasing or convergence-rate analysis would be a natural follow-up.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 8 minor

Summary. The paper proposes rBDLR, an unsupervised representation-learning model that jointly performs latent low-rank representation, adaptive locality-preserving feature extraction, and block-diagonal structuring of the coding coefficients. The model works in a recovered clean data space, uses a Frobenius-norm surrogate for rank, enforces a k-block-diagonal weight matrix W, and pulls the coefficient matrix Z toward W through a flexible bias term. The optimization is carried out by an inexact augmented Lagrange multiplier method with alternating updates. Experiments compare rBDLR with LPP, IsoP, IRPCA, I-LSPFC, LatLRR, rLatLRR, FLLRR, rLRR, BDR, and others on UMIST, COIL100, ETH80, USPS, and AR for recognition, robustness to noise, and clustering, with additional visualizations of coefficient matrices and recovered faces. The authors report consistent improvements over the compared methods and attribute these gains to the joint robust recovery, adaptive weighting, and block-diagonal constraints.

Significance. If the reported results hold, the paper offers a reasonably unified formulation that extends FLLRR and BDR with an interpretable block-diagonal weight matrix and adaptive locality preservation, and it demonstrates competitive performance across several standard benchmarks. The authors also provide useful empirical evidence on parameter sensitivity and convergence behavior, and the visualizations help illustrate the qualitative behavior of the coefficients. The main significance is tempered, however, by three issues: the number of blocks k is taken from ground-truth class labels, so the 'unsupervised subspace recovery' claim is partially label-driven; the recognition tables report no variance or significance tests; and the convergence of the multi-block inexact ALM scheme is not theoretically guaranteed. These issues do not invalidate the empirical claims for applications where k is known, but they do limit the strength of the central discovery claim and the generality of the reported superiority.

major comments (5)
  1. [§3.2 and §5.6] The value of k in the k-block-diagonal regularizer is set to the number of subject classes in all experiments, and in the clustering experiments k is fixed to the chosen number K of categories. This means the block-diagonal structure is injected from ground-truth labels rather than discovered from the data, which weakens the 'unsupervised subspace discovery' claim in the abstract and Section 1. Please report results with k varying around the true value (e.g., k-2, k-1, k+1, k+2) and, ideally, with an estimated k, or explicitly reframe the claim as applying only when the number of blocks is known.
  2. [Eq. (9), Eq. (12), and Fig. 4] Because W is explicitly constrained to be a symmetric nonnegative k-block-diagonal matrix and Z is minimized against ||W - Z - θ1^T||², the block-diagonal appearance of Z in Fig. 4 is imposed by construction rather than discovered from the data. The visualization therefore does not provide independent evidence of subspace recovery. Please include an ablation without the block-diagonal regularizer, or quantify how block-diagonal Z would be if only the adaptive locality term were used, so the reader can separate the contribution of the regularizer from the data-driven structure.
  3. [Tables 1–4] The recognition results report only averaged accuracies over 10 random splits, with no standard deviations or significance tests. Several improvements are small (e.g., Table 3, rBDLR 90.39% vs rLRR 88.44% at 2 train, and 93.55% vs 92.29% at 8 train), so it is unclear whether the differences are statistically meaningful. Please add standard deviations, error bars, or paired tests (e.g., Wilcoxon signed-rank) for the recognition experiments.
  4. [§5.1] Parameter selection is performed on ETH80 only, with the statement that 'similar findings are obtained from other data sets' but without presenting those results. Since α, β, and γ are then used for all datasets, the possibility of overfitting to the validation set is not eliminated. Please report parameter-sensitivity plots or selected ranges for each dataset, or provide a clear protocol for how the parameters would be chosen on unseen data.
  5. [§3.2 and Fig. 3] The authors correctly note that inexact ALM convergence for problems with more than two blocks is not theoretically guaranteed. The empirical convergence results in Fig. 3 average over 10 splits, but the paper does not report whether any splits failed to converge or how the stopping criterion behaves across runs. Please report the number of iterations and any convergence failures per dataset, and include a stationarity measure (e.g., residual norms) for individual runs.
minor comments (8)
  1. [Eq. (10)] The definitions of A+ and A− are typeset in a garbled way; the matrix entries are not clearly separated, which makes the formulation hard to read.
  2. [Eqs. (17)–(19)] Several quantities in the update equations, such as Ψ, Γ, and the symbol N in Eq. (19), are not explicitly defined in the text, forcing the reader to reverse-engineer them from the surrounding equations.
  3. [Algorithm 1] The initialization lines for μ0, μmax, and the convergence tolerance ε are duplicated, which creates confusion about the intended initialization.
  4. [Fig. 5] The label 'Nosied Images' in Figure 5 is a typo and should read 'Noised Images'.
  5. [§5.4] The phrase 'one-Nearest-Neighbor (1NN) classifier' should be 'one-nearest-neighbor (1NN) classifier'.
  6. [References] Reference [6] lists the author as 'S. Cand' rather than the full name, which is inconsistent with the other references.
  7. [Table 6] The table caption uses a Chinese colon 'Table 6：' instead of an English colon; please standardize the formatting.
  8. [§5.4] The paper claims efficiency advantages of the Frobenius-norm formulation over nuclear-norm methods, but no runtime or computational-cost comparisons are reported. Adding timing information would support this claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the block-diagonal structure is an explicit modeling prior, and the reported recognition/clustering gains are evaluated on external datasets.

full rationale

The paper's objective in Eq. (9) explicitly includes a k-block-diagonal regularizer on W and a term ||W - Z - theta 1^T||_F^2 that pulls Z toward W. This is a stated modeling assumption, not a hidden reduction: the paper says it is 'to force the coefficients to be block-diagonal,' so the resulting block-diagonal coefficient matrix is the intended effect of the constraint rather than a prediction derived from unsupported inputs. The empirical claims (recognition and clustering accuracies) are measured on external benchmarks (UMIST, COIL100, ETH80, USPS) with held-out training/test splits, so they do not reduce to the model's own definition. The setting of k to the number of subject classes (Section 3.2, and K in Section 5.6) is a parameter-choice limitation for the 'unsupervised' narrative, since it uses class-count information; however, it does not supply the partition itself, and the k-block-diagonal regularizer is itself taken from the cited external BDR work. No uniqueness theorem or load-bearing self-citation chain is invoked: the many self-citations appear only as related work and are not used to justify the core claim. Hence no circular step that fits the enumerated patterns can be quoted.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests on standard optimization surrogates, a borrowed block-diagonal regularizer, a low-rank subspace model, and design choices in the objective. No new physical or statistical entities are introduced. The main free parameters are the three regularization weights and the number of blocks k.

free parameters (4)
  • alpha
    Weight for the block-diagonal approximation term; tuned by grid search on ETH80, with alpha >= 1e-5 reported as a good range.
  • beta
    Weight for the adaptive locality reconstruction term; tuned by grid search on ETH80, with beta <= 1e-1 reported as a good range.
  • gamma
    Weight for the L2,1 error term; tuned by grid search on ETH80 and reported as relatively insensitive.
  • k = number of classes per dataset
    Order of the k-block-diagonal regularizer; set to the number of true subject classes, which uses label information in an otherwise unsupervised method.
assumptions (5)
  • standard math The Frobenius norm is a valid convex surrogate for rank in the low-rank representation setting.
    Used throughout Eq. (3) and Eq. (9) to avoid SVD, following FLLRR [19]; not re-derived in this paper.
  • standard math The k-block-diagonal regularizer from BDR [20] is convex and its reformulation with Laplacian and matrix M is valid.
    Borrowed from [20] and applied in Eqs. (22)-(24).
  • domain assumption Data are drawn from a union of k low-rank subspaces and can be decomposed as X = U + E with a sparse error E.
    Core model assumption in Eq. (9); if data do not follow this structure, the recovered clean space and block-diagonal coefficients are not meaningful.
  • domain assumption The adaptive weight matrix W should be symmetric, nonnegative, and have zero diagonal.
    Constraints diag(W)=0, W>=0, W=W^T in Eq. (9), inherited from BDR.
  • ad hoc to paper The flexible bias term theta 1^T in the approximation W approximately Z + theta 1^T is a useful design choice.
    Introduced in Eq. (8) to absorb mixed signs and relax the hard constraint; its exact form is not derived from data.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Robust Subspace Discovery by Block-diagonal Adaptive Locality-constrained Representation." pith.science (2026). https://pith.science/paper/OMY3QK5F

@misc{pith2026190801266,
  author       = {Pith},
  title        = {Pith review of: Robust Subspace Discovery by Block-diagonal Adaptive Locality-constrained Representation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OMY3QK5F}},
  note         = {Machine review of arXiv:1908.01266}
}
read the original abstract

We propose a novel and unsupervised representation learning model, i.e., Robust Block-Diagonal Adaptive Locality-constrained Latent Representation (rBDLR). rBDLR is able to recover multi-subspace structures and extract the adaptive locality-preserving salient features jointly. Leveraging on the Frobenius-norm based latent low-rank representation model, rBDLR jointly learns the coding coefficients and salient features, and improves the results by enhancing the robustness to outliers and errors in given data, preserving local information of salient features adaptively and ensuring the block-diagonal structures of the coefficients. To improve the robustness, we perform the latent representation and adaptive weighting in a recovered clean data space. To force the coefficients to be block-diagonal, we perform auto-weighting by minimizing the reconstruction error based on salient features, constrained using a block-diagonal regularizer. This ensures that a strict block-diagonal weight matrix can be obtained and salient features will possess the adaptive locality preserving ability. By minimizing the difference between the coefficient and weights matrices, we can obtain a block-diagonal coefficients matrix and it can also propagate and exchange useful information between salient features and coefficients. Extensive results demonstrate the superiority of rBDLR over other state-of-the-art methods.

Figures

Figures reproduced from arXiv: 1908.01266 by the authors.

Figure 1
Figure 1. Flow-diagram of our rBDLR, where we show the relations of recovered data, salient features and adaptive coefficients. suffers from several problems to degrade the performance and efficiency. First, LatLRR uses the Nuclear-norm to approximate the rank function for low-rank coding, but the Nuclear-norm based optimization involves the Singular Value Decomposition (SVD) of matrices, which is usually time-consuming espec… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 49 canonical work pages

  1. [41]

    Zhang, W

    Z. Zhang, W. Jiang, Z. Zhang, S. Li, G. Liu and J. Qin (2019). Scalable Block - Diagonal Locality -Constrained Projective Dictionary Learning, In: Proceedings of IJCAI, Macao, China

    work page 2019

  2. [1]

    J. Wu, A. Guo, V. S. Sheng, P. Zhao, Z. Cui, H. Li (2017). Adaptive Low- Rank Multi-Label Active Learning for Image Classification, In: Proceedings of ACM Multimedia, 1336-1344

    work page 2017

  3. [2]

    Qin, J.Yan, K

    Z. Qin, J.Yan, K. Ren, C. Chen, and C. Wang ( 2014). Towards Efficient Privacy-preserving Image Feature Extraction in Cloud Computing, In: Proceedings of ACM Multimedia, 497-506

    work page 2014

  4. [3]

    Zhang, Y

    Z. Zhang, Y. Zhang, G. Liu, J. Tang, S. Yan and M. Wang (2019). Joint Label Prediction based Semi-Supervised Adaptive Concept Factorization for Robust Data Representation, IEEE Trans. on Knowledge and Data Engineering (Early Access). DOI: 10.1109/TKDE.2019.2893956

    arXiv 2019

  5. [4]

    Y. Wang, X. Lin, L. Wu, W. Zhang, Q. Zhang (2014). Exploiting Correlation Consensus: Towards Subspace Clustering for Multi -modal Data, In: Proceedings of ACM Multimedia, 981-984

    work page 2014

  6. [5]

    Zhang, M

    Z. Zhang, M. Zhao, F. Li, L. Zhang and S. Yan (2017). Robust Alternating Low-Rank Representation by Joint Lp - and L2,p-norm Minimization, Neural Networks, 96, 55-70

    work page 2017

  7. [6]

    Cand, and X

    S. Cand, and X. Li (2 009). Robust principal component analysis?, Journal of the ACM, 58(3), 11:1-11:37

    work page

  8. [7]

    Rao (2009)

    J.Wright, A.Ganes and S. Rao (2009). Robust Principal Component Analysis: Exact Recovery of Corrupted Low-Rank Matrices via convex optimization, In: Proceedings of NIPS, 2080-2088

    work page 2009

Show all 50 references
  1. [8]

    B. K. Bao, G. Liu, and C. Xu (2012). Inductive Robust Principal Component Analysis, IEEE Transactions on Image Processing, 21(8), 3794-3800

    2012

  2. [9]

    G. Liu, Z. Lin, Y. Yu (2010). Robust Subspace Segmentation by Low -Rank Representation, In: Proceedings of ICML, 663-670

    2010

  3. [10]

    G. Liu, Z. Lin, and S. C. Yan (2013). Robust Recovery of Subspace Structures by Low -Rank Representation, IEEE Trans. Pattern Anal. Mach. Intell. , 35(1), 171-184

    2013

  4. [11]

    G. C. Liu, H. Xu, J. H. Tang, Q. S. Liu, and S. C. Yan (2016). A Determini stic Analysis for LRR, IEEE Trans. Pattern Anal. Mach. Intell., 38(3), 417-430

    2016

  5. [12]

    T. Li, B. Cheng, B. B. Ni, G.C. Liu, and S. C. Yan (2016). Multitask Low -Rank Affinity Graph for Image Segmentation and Image Annotation, ACM Transactions on Intelligent Systems and Technology, 7(4), 65:1-65:18

    2016

  6. [13]

    M. Zhan, S. Cao, B. Qian, S. Chang and J. Wei (2016). Low -Rank Sparse Feature Selection for Patient Similarity Learning, In: Proceedings of the International Conf. on Data Mining (ICDM), Barcelona, Spain

    2016

  7. [14]

    M. A. Schmitz, M. Heitz, N. Bonneel, F. Ngole, D. Coeurjolly, M. Cuturi, Gabriel Peyré, J.L. Starck (2018). Wasserstein Dic tionary Learning: Optimal Transport-Based Unsupervised Nonlinear Dictionary Learning, SIAM Journal on Imaging Sciences, 11(1), 643-678

    2018

  8. [15]

    Zhang, W

    Z. Zhang, W. Jiang, J. Qin, L. Zhang, F. Li, M. Zhang and S . C. Yan (2018). Jointly Learning Structured Analysis Discri minative Dictionary and Analysis Multiclass Classifier, IEEE Transactions on Neural Networks and Learning Systems, 29(8), 3798-3814

    2018

  9. [16]

    G. C. Liu, and S. C. Yan (2011). Latent low -rank representation for subspace segmentation and feature extraction, In: Proceedings of ICCV, 1615-1622

    2011

  10. [17]

    Zhang, S

    Z. Zhang, S. C. Yan, and M. B. Zhao (2014). Similarity preserving low -rank representation for enhanced data representation and effective subspace learning, Neural Networks, 53, 81-94

    2014

  11. [18]

    Zhang, F

    Z. Zhang, F. Li, M. Zhao, L. Zhang and S. C. Yan (2016). Joint low -rank and sparse principal feature coding for enhanced robust representation and visual classification, IEEE Trans. on Image Proc., 25(6), 2429-2443

    2016

  12. [19]

    Yu and Y

    S. Yu and Y. Q. Wu (2018). Subspace clustering based on latent low rank representation with Frobenius norm minimization, Neurocomputing, 275, 2479-2489

    2018

  13. [20]

    C. Lu, J. S. Feng, Z. C. Lin, T. Mei, and S. C. Yan (2019). Subspace clustering by block diagonal representation, IEEE Trans. Pattern Anal. Mach. Intell., 41(2), 487-501

    2019

  14. [21]

    Y. Yang, H . T. Shen, Z. G. Ma, Z. Huang, and X. F. Zhou (2011). L2, 1 -Norm Regularized Discriminative Feature Selection for Unsupervised Learning, In: Proceeding of IJCAI, 22(1), 1589

    2011

  15. [22]

    Z. Lin, M. Chen, L. Wu, and Y. Ma (2009). The augmented Lagrange multiplier method for exact recovery of corrupted low -rank matrices, Tech. Tep. UILU-ENG-09-2215

    2009

  16. [23]

    Graham and N

    D. Graham and N. Allinson (1998). Characterizing virtual eigensignatures for general purpose face recognition, in Face Recognition: From Theory to Application (NATO ASI Ser ies F, Computer and Systems Sciences) , Springer, Berlin, Heidelberg, 163, 446-456

    1998

  17. [24]

    S. A. Nene, S. K. Nayar , H. Murase (1996). Columbia Object Image Library (COIL-100), Technical Report CUCS-006-96

    1996

  18. [25]

    Leibe, B

    B. Leibe, B. Schiele (2003). Analyzing apperance an d contour based methods for object categorization, Proceeding of CVPR, Madison, WI, USA

    2003

  19. [26]

    Zhang, Z

    H. Zhang, Z. C. Lin, C. Zhang and J. Gao (2014). Robust latent low rank representation for subspace clustering, Neurocomputing, 145, 369-373

    2014

  20. [27]

    Hull (1994)

    J. Hull (1994). A datab ase for handwritten text recognition research, IEEE Transactions on pattern analysis and machine intelligence, 16(5), 550-554

    1994

  21. [28]

    He and P

    X. He and P. Niyogi (2003). Locality preserving projections, In: Proceedings of Neural Information Processing System (NIPS), 16

    2003

  22. [29]

    Han (2007)

    D.Cai, X.He, and J. Han (2007). Isometric projection, In: Proceedings of AAAI Conference on Artificial Intelligence, Vancouver Canada

    2007

  23. [30]

    Jolliffe (2002)

    I. Jolliffe (2002). Principal Component Analysis, second edition, Springer - Verlag, New York

    2002

  24. [31]

    A.Martinez and R.Benavent e (1998), The AR Face Database, CVC Technical Report

    1998

  25. [32]

    H. Wang, Y. Chen, and C. Chiang (2016). Discriminative Paired Dictionary Learning for Visual Recognition, In: Proceedings of ACM Multimedia, 67-71

    2016

  26. [33]

    Y. Lee, C. Wang, S. Mathulaprangsan, J. Zhao, and J. Wang (2016). Locality - preserving K-SVD Based Joint Dictionary and Classifier Learning for Object Recognition, In: Proceedings of ACM Multimedia, 481-485

    2016

  27. [34]

    Z. Liu, J. Wang, G. Liu, J. Pu (2019). Sparse Low -Rank Preserving Projection for Dimensionality Reduction, IEEE Access

    2019

  28. [35]

    Q. Li, Y. Liu, S. Wang, Q. Gao and X. Gao (2019). Image Classification Using Low-Rank Regularized Extreme Learning Machine, IEEE Access, 7, 877-883

    2019

  29. [36]

    S. Zhan, J. Wu, N. Han, J. Wen, X. Fang (2019). Unsupervised feature extra - ction by l ow-rank and sparsity preserving embedding, Neural Networks, 109, 56-66

    2019

  30. [37]

    L. Wang, B. Wang, Z. Zhang, Q. Ye, L. Fu, G. Liu , M. Wang (2019). Robust Auto-weighted Projective Low -Rank and Sparse Recovery for Visual Representation, Neural Networks (NN), 117, 201-215

    2019

  31. [38]

    G. Liu, Z. Zhang, Q. Liu and H. Xiong (2019). Robust Subspace Clustering with Compressed Data, IEEE Trans. Image Processing (Early Access)

    2019

  32. [39]

    Y. Chen, Z. Lai, W. K. Wong, L. Shen, Q. Hu (2018). Low -Rank Linear Embedding for Image Recognition, IEEE Transactions on Multimedia, 20(12), 3212-3222

    2018

  33. [40]

    Zhang, F

    Z. Zhang, F. Li, M. Zhao, L. Zhang and S. Yan (2017). Robust Neighborhood Preserving Projection by Nuclear/L2,1 -Norm Regularization for Image Feature Extraction, IEEE Transactions on Image Processing, 26(4), 1607-1622

    2017

  34. [42]

    Zhang, Y

    Z. Zhang, Y. Zhang, S. Li, G. C. Liu, M. Wang, and S. C. Yan (2019). Robust Unsupervised Flexible Auto -weighted Local -Coordinate Concept Factorization for Image Clustering, In: Proceedings of ICASSP, Brighton, UK

    2019

  35. [43]

    Z. Ding, Y. Fu (2019). Dual Low-Rank Decompositions for Robust Cross-View Learning, IEEE Trans. Image Processing, 28(1), 194-204

    2019

  36. [44]

    Y. Wang, L . Wu, X. Lin, J . Gao (2018). Multi -View Spectral Clustering via Structured Low -Rank Matrix Factorization. IEEE Trans. Neur Netw. Learning Syst, 29(10), 4833-4843

    2018

  37. [45]

    L. Wang, Z. Zhang, S. Li, G. Liu, C. Hou and J. Qin (2018) Similarity-Adaptive Latent Low -Rank Representation for Robust Data Representation , In: Proceedings of the Pacific Rim International Conference on Artificial Intelligence (PRICAI), Nanjing, China

    2018

  38. [46]

    L. Wang, Z. Zhang, G. Liu, Q. Ye, . Qin, and M. Wang (2018). Robust Adaptive Low-Rank and Sparse Embedding for Feature Representation," In: Proceedings of the International Conference on Pattern Recognition (ICPR), Beijing, China

    2018

  39. [47]

    Wang and L

    Y. Wang and L. Wu (2018). Beyond Low -Rank Representation: Orthogonal Clustering Basis Reconstruction with Optimized Graph Structure for Multi - view Spectral Clustering, Neural Networks, 103, 1-8

    2018

  40. [48]

    Lezama, Q

    J. Lezama, Q. Qiu, P. Musé, and G. Sapiro (2019). OLÉ: Orthogonal Low -Rank Embedding - A Plug and Play Geometric Loss for Deep Learning, In: Proceedings of CVPR, 8109-8118

    2019

  41. [49]

    Zhang, X

    X. Zhang, X. Yuan, and L. Carin (2018). Nonlocal Low-Rank Tensor Factor Analysis for Image Restoration, In: Proceedings of CVPR, 8232-8241

    2018

  42. [50]

    Han and J

    B. Han and J. Sim (2018). Reflection Removal Using Low -Rank Matrix Completion, In: Proceedings of CVPR, 3872-3880

    2018

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.