Transformer-Guided Content-Adaptive Graph Learning for Hyperspectral Unmixing

Hui Chen; Liangyu Liu; Wanquan Liu; Xianchao Xiu

REVIEW 4 major objections 4 minor 16 references

Transformer-Guided Content-Adaptive Graph Learning for Hyperspectral Unmixing

T0 review · 4 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read Hybrid transformer-graph net tops hyperspectral unmixing tests

desk verdict Plausible architecture, thin evaluation: the SOTA claim rests on a single run with beta tuned on the test scene. read the letter →

arxiv 2509.03376 v2 pith:WWVP2S4L submitted 2025-09-03 cs.CV

Hui Chen , Liangyu Liu , Xianchao Xiu , Wanquan Liu This is my paper · ORCID

classification cs.CV

keywords hyperspectralunmixingtransformergraphneuralnetworkcontent-adaptivelinearmixingmodelabundanceestimationspectralangledistanceremotesensing

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 tries to establish that hyperspectral unmixing is best served by a network that models both global spectral-spatial dependencies and local consistency at the same time. It proposes T-CAGU, which lets a transformer attend over spectral and spatial tokens and then refines the resulting abundances with a graph whose edges are rebuilt from the transformer's own features. The graph uses several propagation orders with learned weights and injects the refined signal back as a residual, so local detail is sharpened without erasing global context. On synthetic scenes with different noise levels and on two real subimages, the method reports lower spectral angle and reconstruction error than four recent deep unmixing baselines. If the claim holds, the practical message is that combining a global transformer prior with an adaptive local graph is a simple recipe for more accurate abundance maps.

What carries the argument

The load-bearing mechanism is the content-adaptive graph residual module: pixels as nodes, edges defined by spectral similarity and spatial proximity on transformer features, symmetric normalization, K-step propagation, and a learned convex combination of propagation orders, followed by residual injection X' = X + beta Y into the decoder. The learnable alpha weights let the network choose its effective receptive field per content, while beta controls how much local refinement is added to the global transformer output. This is what the paper credits for noise robustness and boundary preservation.

What would settle it

Train T-CAGU on one subimage and evaluate on a separate, unseen hyperspectral scene of the same sensor without re-tuning beta or alpha; if its SAD/RMSE advantage over the same baselines shrinks or disappears, the reported superiority is scene-specific fitting rather than a general unmixing improvement.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that global attention and local graph propagation are complementary rather than competing inductive biases for unmixing. The transformer branch produces a feature sequence in which the class token of the spatial branch is prepended to the spectral sequence and vice versa, forcing each modality to carry global context from the other. The graph then treats every pixel as a node, computes edge weights from both spectral similarity and spatial proximity of transformer features, normalizes the adjacency, and performs K propagation steps whose outputs are combined with learned non-negative weights. Adding a scaled graph residual to the transformer

Load-bearing premise

The evaluation assumes that training and testing on the same small scene, and selecting the residual-strength parameter on that same scene, measures true unmixing ability rather than fitting that one image.

Editorial extensions

If this is right

If true, autoencoder-style unmixing networks should no longer choose between global attention and local smoothing; the two can be composed via residual graph refinement.
Because edge weights are rebuilt from learned features, the graph adapts to content during training, which should make unmixing less sensitive to noise than static spectral-similarity graphs.
The learned multi-order propagation weights provide a data-driven receptive field, potentially reducing hand-tuning of neighborhood sizes.
On the reported real datasets, T-CAGU yields the best mean spectral angle distance, suggesting clearer abundance maps for three- and four-endmember scenes.

Reading between the lines

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

The paper fixes beta and the graph weights per dataset; a natural extension it does not test is making both input-dependent so each image selects its own global-local balance during inference.
The residual graph can be read as a learned smoother whose effective bandwidth is set by the alpha weights; if that reading is right, the learned alpha values on noisy versus clean scenes would show how much smoothing the network actually chooses.
The same global-attention-plus-adaptive-graph recipe could transfer to other per-pixel prediction tasks with sharp boundaries, but the paper does not claim or test that.
The transformer branch exchanges class tokens across spectral and spatial streams; the paper does not ablate this cross-guidance alone, so a direct test would isolate whether it, rather than the graph, drives the global-dependency gain.

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.

Desk Editor's Note

read the letter

Colleague, quick take on arXiv:2509.03376. The T-CAGU framework combines a transformer with cross-branch class tokens, content-adaptive multi-order graph propagation, and residual injection. That specific combination is new relative to the cited baselines, and the architecture is described coherently. The ablation (Case I vs III) suggests the dynamic graph helps, and the code is public. So there is a real kernel here.

The soft spot is the experimental support. Tables I–III report single runs, no error bars. On Samson, beta=0.2 is picked from a sensitivity sweep on the same 95×95 subimage used for the comparison, and SAD/RMSE are exactly the training losses. That makes the reported 0.0343 mean SAD a fit statistic, not necessarily a generalization measure. The stress-test note is right: the method's edge over SSAF-Net on Samson could be selection or seed luck. Jasper is more credible since beta is not re-tuned there, but still single-run. The synthetic table has a suspicious outlier: SAD 0.0092 at SNR=40 vs 0.0584 at SNR=30, with similar RMSE; that needs a recheck. Also Eq. (22) looks like a typo: the SAD sum runs over P endmembers rather than N pixels. And they never compare against their own multi-order graph-regularized NMF from ref [15], which is the closest prior.

None of this invalidates the architecture, but it does mean the 'state-of-the-art' claim is not yet supported. A serious revision should add repeated runs with seeds, a validation-split beta selection or at least a beta sweep on a separate scene, error bars, and the [15] comparison. For a four-page letter, the bar can be lower, but the SOTA claim needs more than one lucky run.

Who is this for? Someone working on hyperspectral unmixing baselines or on transformer/graph hybrids for remote sensing. It is a plausible incremental method, not a conceptual shift. I would not cite it in my own work yet, but I would read a revised version.

Recommendation: send to peer review. The architecture is coherent, the code is available, and the weaknesses are fixable with more careful experiments. With proper error bars and beta handling, it could be a solid contribution.

Referee Report

4 major / 4 minor

Summary. The manuscript proposes T-CAGU, a hyperspectral unmixing framework that combines a transformer branch for global dependencies with a content-adaptive graph module for local consistency. The encoder extracts spectral-spatial features, a transformer with cross-branch class tokens produces global representations, and a multi-order graph propagation with learnable weights refines abundances in a residual manner. The decoder estimates endmembers and abundances under sum-to-one constraints. Experiments are reported on one synthetic dataset and two real subimages (Samson, Jasper Ridge), comparing against DeepTrans, A2SAN, DFNN, and SSAF-Net. The paper claims state-of-the-art SAD/RMSE on synthetic data and best mean SAD on both real datasets.

Significance. If the results hold, the architecture would be a useful contribution to hyperspectral unmixing: it is technically coherent, the idea of using transformer features to build a dynamic graph and fuse multiple propagation orders is reasonable, and the code is promised to be public. The main contribution is empirical, however, and the current evidence is not yet convincing. The manuscript would be strengthened by repeated-seed experiments, validation-based hyperparameter selection, and correction of the loss formulation. As presented, the claim of state-of-the-art performance rests on a small set of single-run comparisons where the main hyperparameter on one real dataset is tuned on the test scene itself.

major comments (4)

[§II-D, Eq. (22)] The SAD loss is written as (1/P) Σ_{i=1}^P arccos(⟨I_i, Î_i⟩/(||I_i||₂||Î_i||₂)). Since P denotes the number of endmembers and I_i, Î_i are spectral vectors, the summation is over the wrong index; it should run over the N pixels (or H·W). This is not a purely cosmetic issue because Eqs. (20)–(22) are both the training objectives and the evaluation metrics. As written, the loss is not reproducible and the reported SAD values cannot be traced to this formula.
[§III-B and §III-D, Table II, Fig. 5] The Samson result uses β=0.2, which is selected by a sensitivity sweep on the same 95×95 subimage that is later used for the comparison in Table II. With a single run per method and no validation split, the reported mean SAD advantage of T-CAGU (0.0343 vs 0.0593 for DeepTrans) may reflect test-set hyperparameter selection or initialization rather than an architectural improvement. The authors should either fix β a priori on a separate validation portion, report results over multiple random seeds with mean±std, or otherwise demonstrate that the β=0.2 choice is not cherry-picked for this particular scene.
[Table I, SNR=40 row] The T-CAGU SAD at SNR=40 is reported as 0.0092, while at SNR=30 it is 0.0584, but the RMSE values are almost unchanged (0.1071 vs 0.1115). Such a non-monotonic jump in SAD with essentially flat RMSE is suspicious and inconsistent with the general trend in the table. This outlier should be rechecked; if it is not a typo, it suggests an instability that would need to be explained. Also, no standard deviations or repeated runs are reported for any table, making it hard to judge whether the differences are meaningful.
[§III-B, Tables II and III] Real-data evaluation is limited to SAD on a single subimage per dataset. No RMSE, abundance accuracy, or statistical significance measures are reported for the real scenes. Since the method is evaluated on the same image used for training, the reported numbers are primarily reconstruction fit on the training image. To support the state-of-the-art claim, the authors should at minimum provide multiple random initializations with error bars, and ideally evaluate on a held-out subimage or with a validation-based early stopping / hyperparameter selection procedure.

minor comments (4)

[Fig. 1] The figure legend contains a typo: 'Abudundance' should be 'Abundance'.
[Eq. (13)] The first exponential uses σ_f in the denominator without squaring, while the second uses σ_g². This is inconsistent and the dimensions of the scaling parameters should be clarified.
[§II-A] The symbol m is used both for the patch block size and for the number of spatial tokens in Eq. (6). This is confusing and should be clarified, for example by using M for the number of spatial tokens.
[§III-C] The ablation study in Fig. 4 reports numerical values but does not describe how the 'static grid graph' is constructed or how the graph module is disabled in Case I. A brief implementation detail would help reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's architecture is empirical, the loss/evaluation metrics are not identical, and the only self-citation is non-load-bearing.

full rationale

This is an empirical architecture paper rather than a derivation from first principles, and I find no step in which a claimed prediction is equivalent by construction to the paper's own inputs. The training loss (Eqs. 20–22) is a reconstruction objective (MSE plus spectral angle between input and reconstructed pixels), whereas the reported SAD/RMSE tables are standard endmember/abundance errors (e.g., Table II reports per-endmember SAD for Soil/Tree/Water); they are related but not identical, so the reader's concern that 'the reported SAD and RMSE are essentially the training objectives' is not supported by the paper's equations. The only self-citation, [15], is background on multi-order graph-regularized NMF and is not load-bearing; no uniqueness theorem or ansatz is imported from the authors' own prior work. The β=0.2 selection on the Samson scene in Section III-D is a test-set hyperparameter choice and a legitimate experimental-validity concern, but it is not a circular derivation: the reported SAD is a measured output of a trained model at that β, not a quantity that is defined as the selection criterion itself. The central claims retain independent empirical content: the synthetic results (Table I) use a fixed β=0.3, and the Jasper results (Table III) are not re-tuned for β on that scene. Thus no circularity score above 0 is warranted.

Assumptions & free parameters 7 free parameters · 4 assumptions · 0 invented entities

The method is an empirical deep-learning pipeline; its central claim rests on many hand-picked hyperparameters (channel counts, token dimensions, patch size, number of propagation steps, edge scaling, residual strength) and on domain assumptions (linear mixing, abundance simplex, validity of training on the target scene). No new physical entities are introduced.

free parameters (7)

Residual strength beta = 0.3 (synthetic), 0.2 (Samson, tuned), unspecified (Jasper)
Controls contribution of graph propagation residual in Eq. (19); selected by sensitivity sweep on the Samson test scene (Section III-D).
Compressed channel count C
Output channels after spectral compression (Section II-A); value not reported.
Patch block size m
Size of spatial block used to extract tokens (Section II-A); not specified.
Number of graph propagation orders K
Number of propagation steps in Eq. (17); not reported.
Token feature dimensions D, S
Dimensionalities of spectral and spatial feature vectors (Section II-A); not reported.
Edge scaling parameters sigma_f, sigma_g
Scaling parameters in Eq. (13); values not reported.
Learnable propagation weights alpha_t = learned during training
Non-negative weights summing to one in Eq. (18); learned during training.

assumptions (4)

domain assumption Linear mixing model I = EM
Assumes each pixel is a linear combination of endmembers with abundances, Eq. (1).
domain assumption Abundance constraints (non-negativity, sum-to-one)
Enforced via softmax over P channels in the decoder (Section II-D).
ad hoc to paper Graph edge weights computed from transformer features are robust to noise
Assumed in Section II-C; the dynamic graph construction relies on this design choice.
ad hoc to paper Training on the same image is a valid evaluation protocol
The experiments train and evaluate on the same scene; no held-out split is used.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Transformer-Guided Content-Adaptive Graph Learning for Hyperspectral Unmixing." pith.science (2026). https://pith.science/paper/WWVP2S4L

@misc{pith2026250903376,
  author       = {Pith},
  title        = {Pith review of: Transformer-Guided Content-Adaptive Graph Learning for Hyperspectral Unmixing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WWVP2S4L}},
  note         = {Machine review of arXiv:2509.03376}
}

read the original abstract

Hyperspectral unmixing (HU) targets to decompose each mixed pixel in remote sensing images into a set of endmembers and their corresponding abundances. Despite significant progress in this field using deep learning, most methods fail to simultaneously characterize global dependencies and local consistency, making it difficult to preserve both long-range interactions and boundary details. This letter proposes a novel transformer-guided content-adaptive graph unmixing framework (T-CAGU), which overcomes these challenges by employing a transformer to capture global dependencies and introducing a content-adaptive graph neural network to enhance local relationships. Unlike previous work, T-CAGU integrates multiple propagation orders to dynamically learn the graph structure, ensuring robustness against noise. Furthermore, T-CAGU leverages a graph residual mechanism to preserve global information and stabilize training. Experimental results demonstrate its superiority over the state-of-the-art methods. Our code is available at https://github.com/xianchaoxiu/T-CAGU.

Figures

Figures reproduced from arXiv: 2509.03376 by the authors.

**Figure 1.** Flowchart of our proposed T-CAGU. The original HSI is processed by a convolutional network to extract compact spectral–spatial features, which [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗

**Figure 2.** Abundance map comparisons on the Samson dataset. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 4.** Ablation studies of the graph module on the Samson dataset. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗

Figures from the paper (1 more)

**Figure 5.** Figure 5: Impact of β on the Samson dataset. [5] B. Sun, Y. Su, H. Sun, J. Bai, P. Li, F. Liu, and D. Liu, “Generative adversarial autoencoder network for anti-shadow hyperspectral unmixing,” IEEE Geoscience and Remote Sensing Letters, vol. 21, pp. 1–5, 2024. [6] J. M. Bioucas-…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 16 canonical work pages

[15]

Adaptive multi-order graph regularized nmf with dual sparsity for hyperspectral unmixing,

H. Chen, L. Liu, X. Xiu, and W. Liu, “Adaptive multi-order graph regularized nmf with dual sparsity for hyperspectral unmixing,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, pp. 1–15, 2025, doi:10.1109/JSTARS.2025.3602505

work page arXiv 2025
[1]

SAAN: Similarity-aware attention flow network for change detection with vhr remote sensing images,

H. Guo, X. Su, C. Wu, B. Du, and L. Zhang, “SAAN: Similarity-aware attention flow network for change detection with vhr remote sensing images,” IEEE Transactions on Image Processing , vol. 33, pp. 2599– 2613, 2024

work page 2024
[2]

A lithological classification model based on fourier neural operators and channel-wise self-attention,

A. Marques, L. J. S. Silva, J. Cagliari, L. Scalco, L. C. F. da Silva, M. R. Veronez, and L. Gonzaga, “A lithological classification model based on fourier neural operators and channel-wise self-attention,” IEEE Geoscience and Remote Sensing Letters , vol. 21, pp. 1–5, 2024

work page 2024
[3]

Variational multiple-instance learning with embedding correlation modeling for hyperspectral target detection,

B. Yang, C. Jiao, J. Wu, and L. Li, “Variational multiple-instance learning with embedding correlation modeling for hyperspectral target detection,” IEEE Transactions on Neural Networks and Learning Sys- tems, vol. 36, no. 8, pp. 13 953–13 967, 2025

work page 2025
[4]

Exploring the potential of enmap hyperspectral data for crop classification: Technique and performance evaluation,

J. Goswami, O. V . Murry, P. Boruah, and S. P. Aggarwal, “Exploring the potential of enmap hyperspectral data for crop classification: Technique and performance evaluation,” ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences , pp. 291–297, 2025. 0.0462 0.0422 0.0343 Case I Case II Case III 0 0.01 0.02 0.03 0.04 0.05 0.06SA...

work page 2025
[5]

Generative ad- versarial autoencoder network for anti-shadow hyperspectral unmixing,

B. Sun, Y . Su, H. Sun, J. Bai, P. Li, F. Liu, and D. Liu, “Generative ad- versarial autoencoder network for anti-shadow hyperspectral unmixing,” IEEE Geoscience and Remote Sensing Letters , vol. 21, pp. 1–5, 2024

work page 2024
[6]

Hyperspectral unmixing overview: Geometrical, statistical, and sparse regression-based approaches,

J. M. Bioucas-Dias, A. Plaza, N. Dobigeon, M. Parente, Q. Du, P. Gader, and J. Chanussot, “Hyperspectral unmixing overview: Geometrical, statistical, and sparse regression-based approaches,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , vol. 5, no. 2, pp. 354–379, 2012

work page 2012
[7]

Hyperspectral sparse unmixing based on dual-population cooperative optimization,

K. Huang, L. Chen, Y . Dong, and Y . Tian, “Hyperspectral sparse unmixing based on dual-population cooperative optimization,” IEEE Geoscience and Remote Sensing Letters , vol. 22, pp. 1–5, 2025

work page 2025

Show all 16 references

[8]

Vertex component analysis: A fast algorithm to unmix hyperspectral data,

J. M. Nascimento and J. M. Dias, “Vertex component analysis: A fast algorithm to unmix hyperspectral data,” IEEE Transactions on Geoscience and Remote Sensing , vol. 43, no. 4, pp. 898–910, 2005

2005
[9]

Hy- perspectral unmixing based on nonnegative matrix factorization: A comprehensive review,

X.-R. Feng, H.-C. Li, R. Wang, Q. Du, X. Jia, and A. Plaza, “Hy- perspectral unmixing based on nonnegative matrix factorization: A comprehensive review,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , vol. 15, pp. 4414–4436, 2022

2022
[10]

Hyper- spectral unmixing using transformer network,

P. Ghosh, S. K. Roy, B. Koirala, B. Rasti, and P. Scheunders, “Hyper- spectral unmixing using transformer network,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–16, 2022

2022
[11]

An abundance-guided attention network for hyperspectral unmixing,

X. Tao, M. E. Paoletti, Z. Wu, J. M. Haut, P. Ren, and A. Plaza, “An abundance-guided attention network for hyperspectral unmixing,” IEEE Transactions on Geoscience and Remote Sensing , vol. 62, pp. 1–14, 2024

2024
[12]

A new dual-feature fu- sion network for enhanced hyperspectral unmixing,

X. Tao, B. Koirala, A. Plaza, and P. Scheunders, “A new dual-feature fu- sion network for enhanced hyperspectral unmixing,” IEEE Transactions on Geoscience and Remote Sensing , vol. 62, pp. 1–13, 2024

2024
[13]

SSAF-Net: A spatial-spectral adaptive fusion network for hyperspectral unmixing with endmember variability,

W. Gao, J. Yang, Y . Zhang, Y . Akoudad, and J. Chen, “SSAF-Net: A spatial-spectral adaptive fusion network for hyperspectral unmixing with endmember variability,” IEEE Transactions on Geoscience and Remote Sensing, vol. 63, pp. 1–15, 2025

2025
[14]

Graph attention convolutional autoencoder-based unsupervised nonlinear unmixing for hyperspectral images,

D. Jin and B. Yang, “Graph attention convolutional autoencoder-based unsupervised nonlinear unmixing for hyperspectral images,” IEEE Jour- nal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 16, pp. 7896–7906, 2023

2023
[16]

A new minimum-volume enclosing algorithm for endmember identification and abundance estimation in hyperspectral data,

E. M. Hendrix, I. Garcia, J. Plaza, G. Martin, and A. Plaza, “A new minimum-volume enclosing algorithm for endmember identification and abundance estimation in hyperspectral data,” IEEE Transactions on Geoscience and Remote Sensing , vol. 50, no. 7, pp. 2744–2757, 2011

2011

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [15]

Adaptive multi-order graph regularized nmf with dual sparsity for hyperspectral unmixing,

H. Chen, L. Liu, X. Xiu, and W. Liu, “Adaptive multi-order graph regularized nmf with dual sparsity for hyperspectral unmixing,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, pp. 1–15, 2025, doi:10.1109/JSTARS.2025.3602505

work page arXiv 2025

[2] [1]

SAAN: Similarity-aware attention flow network for change detection with vhr remote sensing images,

H. Guo, X. Su, C. Wu, B. Du, and L. Zhang, “SAAN: Similarity-aware attention flow network for change detection with vhr remote sensing images,” IEEE Transactions on Image Processing , vol. 33, pp. 2599– 2613, 2024

work page 2024

[3] [2]

A lithological classification model based on fourier neural operators and channel-wise self-attention,

A. Marques, L. J. S. Silva, J. Cagliari, L. Scalco, L. C. F. da Silva, M. R. Veronez, and L. Gonzaga, “A lithological classification model based on fourier neural operators and channel-wise self-attention,” IEEE Geoscience and Remote Sensing Letters , vol. 21, pp. 1–5, 2024

work page 2024

[4] [3]

Variational multiple-instance learning with embedding correlation modeling for hyperspectral target detection,

B. Yang, C. Jiao, J. Wu, and L. Li, “Variational multiple-instance learning with embedding correlation modeling for hyperspectral target detection,” IEEE Transactions on Neural Networks and Learning Sys- tems, vol. 36, no. 8, pp. 13 953–13 967, 2025

work page 2025

[5] [4]

Exploring the potential of enmap hyperspectral data for crop classification: Technique and performance evaluation,

J. Goswami, O. V . Murry, P. Boruah, and S. P. Aggarwal, “Exploring the potential of enmap hyperspectral data for crop classification: Technique and performance evaluation,” ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences , pp. 291–297, 2025. 0.0462 0.0422 0.0343 Case I Case II Case III 0 0.01 0.02 0.03 0.04 0.05 0.06SA...

work page 2025

[6] [5]

Generative ad- versarial autoencoder network for anti-shadow hyperspectral unmixing,

B. Sun, Y . Su, H. Sun, J. Bai, P. Li, F. Liu, and D. Liu, “Generative ad- versarial autoencoder network for anti-shadow hyperspectral unmixing,” IEEE Geoscience and Remote Sensing Letters , vol. 21, pp. 1–5, 2024

work page 2024

[7] [6]

Hyperspectral unmixing overview: Geometrical, statistical, and sparse regression-based approaches,

J. M. Bioucas-Dias, A. Plaza, N. Dobigeon, M. Parente, Q. Du, P. Gader, and J. Chanussot, “Hyperspectral unmixing overview: Geometrical, statistical, and sparse regression-based approaches,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , vol. 5, no. 2, pp. 354–379, 2012

work page 2012

[8] [7]

Hyperspectral sparse unmixing based on dual-population cooperative optimization,

K. Huang, L. Chen, Y . Dong, and Y . Tian, “Hyperspectral sparse unmixing based on dual-population cooperative optimization,” IEEE Geoscience and Remote Sensing Letters , vol. 22, pp. 1–5, 2025

work page 2025