Pith. sign in

REVIEW 5 major objections 6 minor 82 references

CL-BioGAN: Biologically-Inspired Cross-Domain Continual Learning for Hyperspectral Anomaly Detection

T0 review · 5 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A biologically inspired continual-learning GAN can detect hyperspectral anomalies across multiple scenes in a single model by actively forgetting conflicting background knowledge.

desk verdict The empirical GAN+replay system may be serviceable, but the paper's advertised bio-inspired active-forgetting loss is not what is implemented, and its own ablation data contradict the central claim. read the letter →

arxiv 2505.11796 v1 pith:SZ5LUXTK submitted 2025-05-17 cs.CV

Jianing Wang , Zheng Hua , Wan Zhang , Shengjia Hao , Yuqiong Yao , Maoguo Gong This is my paper · ORCID
classification cs.CV
keywords hyperspectralanomalydetectioncontinuallearningactiveforgettinggenerativeadversarialnetworkself-attentionBayesianreplaystrategybackgroundsuppression
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 claims that a single generative model can keep learning to find anomalies in new hyperspectral scenes, one after another, without losing the ability to detect anomalies in scenes it has already seen. The proposed model, CL-BioGAN, couples a bio-inspired active-forgetting loss with a replay buffer and a self-attention GAN that reconstructs background pixels. If the claim holds, continual learning would let hyperspectral anomaly detectors be updated scene by scene with far fewer parameters than retraining or than the previous capsule-based continual GAN, while keeping accuracy stable.

What carries the argument

The load-bearing object is the CL-Bio Loss, which is the sum of an Active Forgetting (AF) loss and a Continual Learning (CL) loss. The AF loss is derived from a Bayesian posterior whose prior is an interpolated, forgetting-factor-weakened version of the old posterior, encouraging the network to release parameters that conflict with the new task; the CL loss is an F-norm regularization on the generator's replay-set covariances that keeps new updates close to old behavior. Around this loss, the BioGAN uses a background selection module with a spectral-angle threshold to pick clean pixels, an encoder-decoder generator, and a discriminator with a local-global self-attention block whose L2-normalized attention is intended to capture long-range spectral correlations. At test time the anomaly map is simply the per-pixel L2 reconstruction error between the input scene and the generator's output.

What would settle it

Retrain CL-BioGAN on the same ABU and HAD100 task sequences with mu set to 0.90, 0.95, 0.99, and 0.999 and compare ACC and BWT; if the reported advantage shrinks or inverts over that range, the fixed threshold is carrying the result. A complementary test is to inject a spectrally subtle synthetic anomaly into the background set B_t and check whether the final anomaly map suppresses it.

Watch

Extended reading notes

Core claim

The central claim is that a continual-learning GAN can fit the changing background distribution of successive hyperspectral scenes and still remember earlier scenes, by actively forgetting the part of old knowledge that conflicts with the new scene. The paper reports that after training on five ABU scenes in sequence, CL-BioGAN reaches the highest average detection accuracy among the compared continual methods for tasks 3, 4, and 5, with near-zero backward transfer, and that it also outperforms CL-CaGAN on the five HAD100 scenes. It further claims this is achieved with fewer training and detection parameters than the capsule-based baseline, so the stability gain is not bought with model size.

Load-bearing premise

The method assumes that every pixel with spectral angle at least 0.99 to its neighbors is clean background, and this fixed threshold is never varied or sensitivity-tested.

Editorial extensions

If this is right

  • A single CL-BioGAN model can be trained sequentially on at least five different hyperspectral scenes and still detect anomalies in earlier scenes, with backward transfer near zero in the reported runs.
  • The active-forgetting loss and replay buffer reduce the number of trainable parameters compared with the capsule-based continual GAN baseline, so the stability-plasticity balance is not bought with larger models.
  • Detection accuracy on the hardest five-task sequences is reported as the highest among the compared methods, and background suppression stays stable as new scenes arrive.
  • The replay strategy needs only tens to a few hundred representative pixels per scene, which keeps the memory cost of rehearsal low.

Reading between the lines

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

  • Editorial inference: the fixed spectral-angle threshold mu=0.99 is the pipeline's most exposed assumption; varying it per scene or learning it from data would likely change the accuracy numbers, and the paper does not report that sensitivity.
  • Editorial inference: the same Bayesian active-forgetting loss could be grafted onto other reconstruction-based anomaly detectors, not only GANs, since the loss acts on network parameters rather than on the GAN machinery.
  • Editorial inference: because replay uses so few pixels, the method might extend to longer task streams, but the paper only demonstrates five consecutive scenes, so scaling behavior beyond that is untested.
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 / 6 minor

Summary. The paper proposes CL-BioGAN, a continual-learning generative adversarial network for cross-scene hyperspectral anomaly detection (HAD). The method combines a spectral-angle background selection module (BSM), K-means-based sample replay, an encoder-decoder generator with a self-attention discriminator, and a Bio-Inspired Loss composed of an Active Forgetting (AF) loss and a CL loss. The AF loss is derived from a Bayesian continual-learning perspective using a forgetting factor beta, while the CL loss is intended to preserve stability via replay. Experiments on the ABU and HAD100 datasets, each with five tasks, compare against joint training, fine-tuning, MAS, EWC, OWM, and CL-CaGAN, reporting ACC, BWT, FWT, and AUC_BS, together with parameter counts.

Significance. If the biological active-forgetting mechanism functioned as described, the paper would offer a novel way to balance stability and plasticity in hyperspectral continual learning. The evaluation is grounded in public benchmarks, includes several continual-learning baselines, and reports backward/forward transfer in addition to average accuracy. The paper also provides a component ablation and parameter-count comparisons, which are useful for reproducibility. However, the central theoretical derivation is circular, the implemented AF loss reduces to ordinary weight decay with no dependence on the forgetting factor, the algorithm pseudocode omits the proposed losses, and the paper's own ablations show that removing the AF loss changes ACC by at most 0.0038 on tasks 3–5. The claimed bio-inspired mechanism is therefore not supported by the evidence presented.

major comments (5)
  1. [Section III-C.2.a, Eq. (29)] The MAP objective for the updated parameters contains log p(theta | X_{t+1}) on the right-hand side, which is the posterior of the new task that the MAP estimate is supposed to define; the derivation is circular. Additionally, the text following Eq. (26) discusses limits of p(theta) rather than the forgetting factor beta: the statement 'When p(theta)→0, ^p will be dominated by p(theta|X_t)' should refer to beta→0, and 'p(theta)→1' should refer to beta→1.
  2. [Section III-C.2.a, Eq. (30) and Algorithm 1] The implemented AF loss, L_AF = lambda_AF * sum_c (theta_c)^2, is plain L2 weight decay; the forgetting factor beta from Eqs. (26)–(28) appears nowhere in the loss, in Algorithm 1, or in the experimental settings. Moreover, Algorithm 1's update steps (lines 8 and 10) minimize only L_G, L_D, and L_recon, so neither the AF loss nor the CL loss is included in the described training procedure, contradicting the claim that these losses are central to the method.
  3. [Section III-C.2.b, Eq. (31)] The CL loss is not reproducible as written: Bb and P are undefined, the norm is called 'Fibonacci norm' (presumably Frobenius), and the expression mixes tensors of unspecified shape. Without definitions of these quantities, the equation cannot be implemented or checked.
  4. [Table IX and Section IV-C.d] The ablation results contradict the paper's claim that the AF loss is a key stabilizer. Removing the AF loss changes ACC by at most 0.0038 on tasks 3–5 (e.g., task 3: 0.9709 vs 0.9734; task 4: 0.9750 vs 0.9788; task 5: 0.9599 vs 0.9602), and on task 2 the variant without the AF loss is slightly better. Thus, the reported accuracy gains cannot be attributed to the proposed bio-inspired loss.
  5. [Section III-A.1, Eqs. (6)–(8)] The BSM threshold mu=0.99 is fixed without any sensitivity analysis. If anomalies leak into the estimated background set B_t, the generator will learn to reconstruct them and the anomaly map from Eq. (32) will miss them. The paper never varies mu or reports sensitivity to this threshold, so the central detection result depends on an unvalidated fixed value.
minor comments (6)
  1. [Eq. (31)] The term 'Fibonacci norm' should be 'Frobenius norm'.
  2. [Algorithm 1] The parameter list includes lambda_AF=0.1 and lambda_CL=0.9, but these hyperparameters are never used in the pseudocode; the pseudocode should be updated to match the actual training objective used in the experiments.
  3. [Eqs. (14)–(15)] The self-attention formula softmax(||Q||_2 ||K||_2^T / sqrt(2C)) V is nonstandard; typically attention uses Q K^T scaled by a dimension factor. The notation should be clarified to avoid confusion.
  4. [Eq. (18)] The replay-selection equation is typeset in a garbled way, with the 'KMeans' function and its indexing unclear; please rewrite this equation and its surrounding explanation more clearly.
  5. [Table III] The header of Table III ("NTotal Selected Total Selected Total Selected") is malformed and the meaning of the final column labeled 'N' is unclear; the table should be reformatted with distinct column headings.
  6. [Section IV-A.3, Eq. (36)] The definition of FWT uses an asterisk superscript AUC*; the text does not specify how AUC* is computed, and the formula as written is not standard. Please clarify.

Circularity Check

3 steps flagged · score 6.0 of 10

AF loss derivation is circular: Eq. 29 contains the posterior it aims to compute; the implemented loss is plain L2 weight decay, and the paper's own ablations show negligible effect, though benchmark comparisons remain independent.

  1. self definitional [Section III-C-2-a, Eq. (29)]
    "Next,p(θ|Xtrain t , Xtrain t+1 ,β ) can also be estimated by the approximated Gaussian using Laplace approximation, and the Maximum-a-Posteriori Estimation (MAP) is: θ∗ t,t+1 = arg max θ logp(θ|Xtrain t , Xtrain t+1 ,β ) = arg max θ (1−β)(logp(Xtrain t+1 |θ) + logp(θ|Xtrain t )) +β logp(θ|Xtrain t+1 ) +const."

    The left-hand side is the MAP estimate of the posterior after tasks t and t+1. The right-hand side contains β log p(θ|Xtrain t+1 ), i.e., the posterior after only the new task. That posterior is exactly the unknown quantity the optimization is supposed to produce, so evaluating the objective in Eq. (29) requires already knowing the solution. The standard Bayesian recursion in the paper's own Eq. (5) uses log p(θ|Xtrain t ) as the prior and contains no p(θ|Xtrain t+1 ) term. Thus Eq. (29) is self-referential by construction, not merely approximate.

  2. renaming known result [Section III-C-2-a, Eq. (30), and Algorithm 1]
    "Then, we obtain the active forgetting AF loss function by synaptic Expansion-Convergence (AF loss). LAF =λAF× X c (θc)2"

    This is ordinary squared-L2 parameter regularization (weight decay). None of the Bayesian active-forgetting ingredients appears: no β, no p(θ|Xtrain t ), no Hessian or Laplace covariance, and no weighted product distribution from Eq. (26). Algorithm 1 and the experimental settings never compute β or the Bayesian quantities. Consequently, the 'bio-inspired AF loss' is a rename of a generic L2 penalty, and the claimed Bayesian derivation in Eqs. (26)-(29) is not the loss that is actually trained.

1 more flagged steps
  1. other [Section IV-C-d, Table IX (ablation w/o LAF)]
    "This performance indicates that AF Loss is a key component for sustaining CL ability and served as a stabilizer for BS."

    Not a circular reduction, but a required missing-support flag: the paper's own Table IX shows w/o LAF ACC of 0.9709/0.9750/0.9599 for tasks 3/4/5 versus 0.9734/0.9788/0.9602 with AF loss (delta at most 0.0038), and on task 2 the w/o-LAF variant scores 0.9766 versus 0.9763, i.e., slightly higher. The text nonetheless calls AF Loss a 'key component' and 'stabilizer,' so the central attribution is contradicted by the manuscript's own ablation numbers.

full rationale

The paper's main quantitative comparisons (Tables IV-VI) are against public ABU and HAD100 benchmarks, so the empirical ranking is not circular: the reported ACC/BWT/FWT values could in principle falsify the method. The circularity is in the theoretical derivation of the Active Forgetting loss. Equation (29) defines the MAP objective for θ* after tasks t and t+1 using β log p(θ|Xtrain t+1 ) on the right-hand side; that is the posterior being estimated, making the equation self-referential by construction. The standard Bayesian recursion in Eq. (5) contains no such term. Moreover, the loss actually implemented and reported in Algorithm 1 and the experimental settings is Eq. (30), L_AF = λ_AF Σ_c θ_c^2, which is plain L2 weight decay; none of the Bayesian machinery (β, the weighted product distribution, Laplace Hessians) appears in the implementation. Table IX further shows that removing L_AF changes ACC by at most 0.0038 on tasks 3-5 and is slightly better on task 2, undercutting the paper's claim that the bio-inspired AF loss is the stabilizer. The CL loss (Eq. 31) is not reproducible as written (Bb and P are undefined, and the norm is called 'Fibonacci norm' rather than Frobenius norm), and λ_AF/λ_CL are tuned on the same ABU tasks (Fig. 8), which risks selection on the test set but is not a derivation-level circularity. Overall, the empirical benchmark results are independent, but the paper's central theoretical novelty reduces to a circular equation plus a renamed L2 penalty, so partial circularity is present.

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

The method introduces no new physical entities. Its load-bearing components are the background selection threshold and the loss weights, several of which are tuned on the evaluation data; the theoretically motivated forgetting factor beta is defined but never used.

free parameters (5)
  • lambda_AF (AF loss weight) = 0.1
    Selected to maximize ACC in Fig. 8 ablation; used in Eq. (30).
  • lambda_CL (CL loss weight) = 0.9
    Selected to maximize ACC in Fig. 8 ablation; used in Eq. (31).
  • mu (SAM background threshold) = 0.99
    Fixed threshold in Eq. (7) to decide background vs anomaly; no sensitivity analysis.
  • P (number of replay clusters) = 3
    Chosen from Table VIII as the best of P=2..6.
  • beta (forgetting factor) = not reported
    Defined in Eq. (26) as the core of the active-forgetting mechanism, said to be grid-searched, but it does not appear in Algorithm 1, the loss equations, or the experiment tables.
assumptions (4)
  • standard math The posterior p(theta|X_t) can be approximated by a Gaussian via Laplace approximation.
    Used in Section II to justify the Bayesian CL formulation, Eqs. (1)-(5).
  • domain assumption Spectral Angle Mapper similarity with threshold mu separates background from anomalies across all scenes.
    Eqs. (6)-(8) assume pixels with SAM < 0.99 are anomalies; if this fails, training data for the GAN is contaminated.
  • domain assumption Background pixels are the overwhelming majority and are reconstructable by an autoencoder/GAN, so reconstruction error identifies anomalies.
    Standard HAD assumption stated in Section III-A-1 and used in Eq. (32).
  • domain assumption The replay set selected by K-Means is representative of old tasks' background distributions.
    Eqs. (18)-(19) assume cluster-center samples preserve old task knowledge.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CL-BioGAN: Biologically-Inspired Cross-Domain Continual Learning for Hyperspectral Anomaly Detection." pith.science (2026). https://pith.science/paper/SZ5LUXTK

@misc{pith2026250511796,
  author       = {Pith},
  title        = {Pith review of: CL-BioGAN: Biologically-Inspired Cross-Domain Continual Learning for Hyperspectral Anomaly Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SZ5LUXTK}},
  note         = {Machine review of arXiv:2505.11796}
}
read the original abstract

Memory stability and learning flexibility in continual learning (CL) is a core challenge for cross-scene Hyperspectral Anomaly Detection (HAD) task. Biological neural networks can actively forget history knowledge that conflicts with the learning of new experiences by regulating learning-triggered synaptic expansion and synaptic convergence. Inspired by this phenomenon, we propose a novel Biologically-Inspired Continual Learning Generative Adversarial Network (CL-BioGAN) for augmenting continuous distribution fitting ability for cross-domain HAD task, where Continual Learning Bio-inspired Loss (CL-Bio Loss) and self-attention Generative Adversarial Network (BioGAN) are incorporated to realize forgetting history knowledge as well as involving replay strategy in the proposed BioGAN. Specifically, a novel Bio-Inspired Loss composed with an Active Forgetting Loss (AF Loss) and a CL loss is designed to realize parameters releasing and enhancing between new task and history tasks from a Bayesian perspective. Meanwhile, BioGAN loss with L2-Norm enhances self-attention (SA) to further balance the stability and flexibility for better fitting background distribution for open scenario HAD (OHAD) tasks. Experiment results underscore that the proposed CL-BioGAN can achieve more robust and satisfying accuracy for cross-domain HAD with fewer parameters and computation cost. This dual contribution not only elevates CL performance but also offers new insights into neural adaptation mechanisms in OHAD task.

Figures

Figures reproduced from arXiv: 2505.11796 by the authors.

Figure 1
Figure 1. Comparison of traditional DL training model and the CL method. (a) [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Principles of active forgetting and learning in our proposed neural network. (a) represents the Neural and Protrusion Function Diagram. (b) represents [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Overview of the proposed CL-BioGAN for OHAD. The left part represents the continual training phase, which mainly composed with Replay [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: The catastrophic forgetting performance evaluation for open scenario OHAD on previous tasks. (a) AUC value for tasks 1-2 after training on the [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: The ACC, BWT and FWT Performance evaluation for OHAD of different methods. (a) The variation of [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: The AUC test results of different methods on the previous task after learning the fifth task are shown as follows: the first column (a) represents the AUC(D,F ) value on Los Angeles-2 after training task 1-5. The second column (b) represents the AUC(D,τ) value on Los A…
Figure 7
Figure 7. Figure 7: Visualization for five OHAD results after learning 1-5 Tasks. The data set from top to down is Los Angeles-1, Cat Island, Los Angeles-2, San Diego, [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: Varied parameter proportions of λAF and λCL for the detection ACC. (a) represents the individual impact of λAF and λCL on the experimental results, while (b) represents the impact of λAF and λCL on the experimental results within a certain range of values. b) Effect an…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

82 extracted references · 60 canonical work pages

  1. [1]

    Dimension- ality reduction via multiple neighborhood-aware nonlinear collaborative analysis for hyperspectral image classification,

    Y . Duan, C. Chen, M. Fu, Y . Li, X. Gong, and F. Luo, “Dimension- ality reduction via multiple neighborhood-aware nonlinear collaborative analysis for hyperspectral image classification,” IEEE Transactions on Circuits and Systems for Video Technology , pp. 1–1, 2024

    2024

  2. [2]

    Hyperspectral anomaly detection for spectral anomaly targets via spatial and spectral constraints,

    Z. Li, Y . Zhang, and J. Zhang, “Hyperspectral anomaly detection for spectral anomaly targets via spatial and spectral constraints,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–15, 2022

    2022

  3. [3]

    One-step detection paradigm for hyperspectral anomaly detection via spectral deviation relationship learning,

    J. Li, X. Wang, S. Wang, H. Zhao, and Y . Zhong, “One-step detection paradigm for hyperspectral anomaly detection via spectral deviation relationship learning,” IEEE Transactions on Geoscience and Remote Sensing, vol. 62, pp. 1–15, 2024

    work page 2024

  4. [4]

    Interactive spectral- spatial transformer for hyperspectral image classification,

    L. Song, Z. Feng, S. Yang, X. Zhang, and L. Jiao, “Interactive spectral- spatial transformer for hyperspectral image classification,” IEEE Trans- actions on Circuits and Systems for Video Technology , pp. 1–1, 2024

    work page 2024

  5. [5]

    Continual lifelong learning with neural networks: A review,

    G. I. Parisi, R. Kemker, J. L. Part, C. Kanan, and S. Wermter, “Continual lifelong learning with neural networks: A review,” Neural Networks, vol. 113, pp. 54–71, 2019

    2019

  6. [6]

    A continual learning survey: Defying forgetting in classification tasks,

    M. De Lange, R. Aljundi, M. Masana, S. Parisot, X. Jia, A. Leonardis, G. Slabaugh, and T. Tuytelaars, “A continual learning survey: Defying forgetting in classification tasks,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 44, no. 7, pp. 3366–3385, 2022

    2022

  7. [7]

    Class-incremental learning: Survey and performance evaluation on image classification,

    M. Masana, X. Liu, B. Twardowski, M. Menta, A. D. Bagdanov, and J. van de Weijer, “Class-incremental learning: Survey and performance evaluation on image classification,” IEEE Transactions on Pattern Anal- ysis and Machine Intelligence , vol. 45, no. 5, pp. 5513–5533, 2023

    work page 2023

  8. [8]

    Catastrophic interference in connec- tionist networks: The sequential learning problem,

    M. McCloskey and N. J. Cohen, “Catastrophic interference in connec- tionist networks: The sequential learning problem,” ser. Psychology of Learning and Motivation. Academic Press, 1989, vol. 24, pp. 109–165

    work page 1989

Show all 82 references
  1. [9]

    Loss of plasticity in deep continual learning,

    S. Dohare, J. F. Hernandez-Garcia, Q. Lan, P. Rahman, A. R. Mahmood, and R. S. Sutton, “Loss of plasticity in deep continual learning,” Nature, vol. 632, no. 8026, pp. 768–774, 2024

    2024

  2. [10]

    A comprehensive survey of continual learning: Theory, method and application,

    L. Wang, X. Zhang, H. Su, and J. Zhu, “A comprehensive survey of continual learning: Theory, method and application,” IEEE Transactions on Pattern Analysis and Machine Intelligence , pp. 1–20, 2024

    2024

  3. [11]

    Esdb: Expand the shrinking decision boundary via one-to-many information matching for continual learning with small memory,

    K. Li, H. Chen, J. Wan, and S. Yu, “Esdb: Expand the shrinking decision boundary via one-to-many information matching for continual learning with small memory,” IEEE Transactions on Circuits and Systems for Video Technology, pp. 1–1, 2024

    2024

  4. [12]

    Incorporating neuro-inspired adaptability for continual learning in artificial intelligence,

    L. Wang, X. Zhang, Q. Li, M. Zhang, H. Su, J. Zhu, and Y . Zhong, “Incorporating neuro-inspired adaptability for continual learning in artificial intelligence,” Nature Machine Intelligence , vol. 5, no. 12, pp. 1356–1368, 2023

    2023

  5. [13]

    AFEC: Active forgetting of negative transfer in continual learning,

    L. Wang, M. Zhang, Z. Jia, Q. Li, C. Bao, K. Ma, J. Zhu, and Y . Zhong, “AFEC: Active forgetting of negative transfer in continual learning,” in Advances in Neural Information Processing Systems , M. Ranzato, A. Beygelzimer, Y . Dauphin, P. Liang, and J. W. Vaughan, Eds., vol....

    2021

  6. [14]

    Synaptic plasticity as bayesian inference,

    L. Aitchison, J. Jegminat, J. A. Menendez, J.-P. Pfister, A. Pouget, and P. E. Latham, “Synaptic plasticity as bayesian inference,” Nature neuroscience, vol. 24, no. 4, pp. 565–571, 2021

    2021

  7. [15]

    Presynaptic stochasticity improves energy efficiency and helps alleviate the stability-plasticity dilemma,

    S. Schug, F. Benzing, and A. Steger, “Presynaptic stochasticity improves energy efficiency and helps alleviate the stability-plasticity dilemma,” eLife, vol. 10, p. e69884, oct 2021

    2021

  8. [18]

    Kaizen: Practical self-supervised continual learning with continual fine-tuning,

    C. I. Tang, L. Qendro, D. Spathis, F. Kawsar, C. Mascolo, and A. Mathur, “Kaizen: Practical self-supervised continual learning with continual fine-tuning,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), January 2024, pp. 2841–2850

    2024

  9. [19]

    Integrating present and past in unsupervised continual learning,

    Y . Zhang, L. Charlin, R. Zemel, and M. Ren, “Integrating present and past in unsupervised continual learning,” arXiv preprint arXiv:2404.19132, 2024

    2024 arXiv

  10. [20]

    Learning prompt with distribution-based feature replay for few-shot class-incremental learning,

    Z. Huang, Z. Chen, Z. Chen, E. Zhou, X. Xu, R. S. M. Goh, Y . Liu, C. Feng, and W. Zuo, “Learning prompt with distribution-based feature replay for few-shot class-incremental learning,” arXiv preprint arXiv:2401.01598, 2024

    2024 arXiv

  11. [21]

    Continual learning of medical image classification based on feature replay,

    X. Li, H. Li, and L. Ma, “Continual learning of medical image classification based on feature replay,” in 2022 16th IEEE International Conference on Signal Processing (ICSP) , vol. 1, 2022, pp. 426–430

    2022

  12. [22]

    Con- trastive continuity on augmentation stability rehearsal for continual self- supervised learning,

    H. Cheng, H. Wen, X. Zhang, H. Qiu, L. Wang, and H. Li, “Con- trastive continuity on augmentation stability rehearsal for continual self- supervised learning,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , October 2023, pp. 5707–5717

    2023

  13. [23]

    BiRT: Bio-inspired replay in vision transformers for continual learning,

    K. Jeeveswaran, P. Bhat, B. Zonooz, and E. Arani, “BiRT: Bio-inspired replay in vision transformers for continual learning,” arXiv preprint arXiv:2305.04769, 2023

    2023 arXiv

  14. [24]

    Exemplar-based contin- ual learning via contrastive learning,

    S. Chen, M. Zhang, J. Zhang, and K. Huang, “Exemplar-based contin- ual learning via contrastive learning,” IEEE Transactions on Artificial Intelligence, vol. 5, no. 7, pp. 3313–3324, 2024

    2024

  15. [25]

    Relational experience replay: Continual learning by adap- tively tuning task-wise relationship,

    Q. Wang, R. Wang, Y . Li, D. Wei, H. Wang, K. Ma, Y . Zheng, and D. Meng, “Relational experience replay: Continual learning by adap- tively tuning task-wise relationship,” IEEE Transactions on Multimedia, pp. 1–15, 2024

    2024

  16. [26]

    Class-incremental learning via deep model consolidation,

    J. Zhang, J. Zhang, S. Ghosh, D. Li, S. Tasci, L. Heck, H. Zhang, and C.- C. J. Kuo, “Class-incremental learning via deep model consolidation,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), March 2020, pp. 1131–1140

    2020

  17. [27]

    Class similarity weighted knowledge distillation for continual semantic segmentation,

    M. H. Phan, T.-A. Ta, S. L. Phung, L. Tran-Thanh, and A. Bouzer- doum, “Class similarity weighted knowledge distillation for continual semantic segmentation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , June 2022, pp. 16 866–16 875

    2022

  18. [28]

    A contrastive continual learning for the classification of remote sensing imagery,

    A. S. Alakooz and N. Ammour, “A contrastive continual learning for the classification of remote sensing imagery,” in IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium , 2022, pp. 7902–7905

    2022

  19. [29]

    Continual barlow twins: Continual self- supervised learning for remote sensing semantic segmentation,

    V . Marsocci and S. Scardapane, “Continual barlow twins: Continual self- supervised learning for remote sensing semantic segmentation,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 16, pp. 5049–5060, 2023

    2023

  20. [30]

    Achieving a better stability-plasticity trade-off via auxiliary networks in continual learning,

    S. Kim, L. Noci, A. Orvieto, and T. Hofmann, “Achieving a better stability-plasticity trade-off via auxiliary networks in continual learning,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2023, pp. 11 930–11 939

    2023

  21. [31]

    Piggyback: Adapting a single network to multiple tasks by learning to mask weights,

    A. Mallya, D. Davis, and S. Lazebnik, “Piggyback: Adapting a single network to multiple tasks by learning to mask weights,” in Proceedings of the European Conference on Computer Vision (ECCV) , September 2018, pp. 67–82

    2018

  22. [32]

    Learn to grow: A continual structure learning framework for overcoming catastrophic forgetting,

    X. Li, Y . Zhou, T. Wu, R. Socher, and C. Xiong, “Learn to grow: A continual structure learning framework for overcoming catastrophic forgetting,” in Proceedings of the 36th International Conference on Machine Learning , ser. Proceedings of Machine Learning Research, K. Chaudh...

    2019

  23. [33]

    Meta-Attention for ViT- backed continual learning,

    M. Xue, H. Zhang, J. Song, and M. Song, “Meta-Attention for ViT- backed continual learning,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , June 2022, pp. 150–159

    2022

  24. [34]

    Adversarial continual learning,

    S. Ebrahimi, F. Meier, R. Calandra, T. Darrell, and M. Rohrbach, “Adversarial continual learning,” in Computer Vision – ECCV 2020 , A. Vedaldi, H. Bischof, T. Brox, and J.-M. Frahm, Eds. Cham: Springer International Publishing, 2020, pp. 386–402

    2020

  25. [35]

    Design of distributed rule-based models in the presence of large data,

    H. E, Y . Cui, W. Pedrycz, and Z. Li, “Design of distributed rule-based models in the presence of large data,” IEEE Transactions on Fuzzy Systems, vol. 31, no. 7, pp. 2479–2486, 2023

    2023

  26. [37]

    Balancing stability and plas- ticity through advanced null space in continual learning,

    Y . Kong, L. Liu, Z. Wang, and D. Tao, “Balancing stability and plas- ticity through advanced null space in continual learning,” in European Conference on Computer Vision . Springer, 2022, pp. 219–236

    2022

  27. [38]

    Embracing change: Continual learning in deep neural networks,

    R. Hadsell, D. Rao, A. A. Rusu, and R. Pascanu, “Embracing change: Continual learning in deep neural networks,” Trends in cognitive sci- ences, vol. 24, no. 12, pp. 1028–1040, 2020

    2020

  28. [39]

    Anti- retroactive interference for lifelong learning,

    R. Wang, Y . Bao, B. Zhang, J. Liu, W. Zhu, and G. Guo, “Anti- retroactive interference for lifelong learning,” in European Conference on Computer Vision . Springer, 2022, pp. 163–178

    2022

  29. [40]

    Optimizing reusable knowledge for continual learning via metalearning,

    J. Hurtado, A. Raymond, and A. Soto, “Optimizing reusable knowledge for continual learning via metalearning,” Advances in Neural Informa- tion Processing Systems , vol. 34, pp. 14 150–14 162, 2021

    2021

  30. [41]

    Towards better plasticity-stability trade-off in incremental learning: A simple linear connector,

    G. Lin, H. Chu, and H. Lai, “Towards better plasticity-stability trade-off in incremental learning: A simple linear connector,” in Proceedings of JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, JULY 2024 15 the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022...

    2024

  31. [42]

    Training networks in null space of feature covariance for continual learning,

    S. Wang, X. Li, J. Sun, and Z. Xu, “Training networks in null space of feature covariance for continual learning,” in Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition , 2021, pp. 184–193

    2021

  32. [43]

    The challenges of con- tinuous self-supervised learning,

    S. Purushwalkam, P. Morgado, and A. Gupta, “The challenges of con- tinuous self-supervised learning,” in European Conference on Computer Vision. Springer, 2022, pp. 702–721

    2022

  33. [44]

    Self-supervised models are continual learners,

    E. Fini, V . G. T. Da Costa, X. Alameda-Pineda, E. Ricci, K. Alahari, and J. Mairal, “Self-supervised models are continual learners,” in Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 9621–9630

    2022

  34. [45]

    Learning to prompt for continual learning,

    Z. Wang, Z. Zhang, C.-Y . Lee, H. Zhang, R. Sun, X. Ren, G. Su, V . Perot, J. Dy, and T. Pfister, “Learning to prompt for continual learning,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 139–149

    2022

  35. [46]

    Coda-prompt: Contin- ual decomposed attention-based prompting for rehearsal-free continual learning,

    J. S. Smith, L. Karlinsky, V . Gutta, P. Cascante-Bonilla, D. Kim, A. Arbelle, R. Panda, R. Feris, and Z. Kira, “Coda-prompt: Contin- ual decomposed attention-based prompting for rehearsal-free continual learning,” in Proceedings of the IEEE/CVF Conference on Computer Vision a...

    2023

  36. [47]

    Dualprompt: Complementary prompting for rehearsal-free continual learning,

    Z. Wang, Z. Zhang, S. Ebrahimi, R. Sun, H. Zhang, C.-Y . Lee, X. Ren, G. Su, V . Perot, J. Dyet al., “Dualprompt: Complementary prompting for rehearsal-free continual learning,” in European Conference on Computer Vision. Springer, 2022, pp. 631–648

    2022

  37. [48]

    S-prompts learning with pre-trained transformers: An occam’s razor for domain incremental learning,

    Y . Wang, Z. Huang, and X. Hong, “S-prompts learning with pre-trained transformers: An occam’s razor for domain incremental learning,” Ad- vances in Neural Information Processing Systems , vol. 35, pp. 5682– 5695, 2022

    2022

  38. [49]

    Passnet: A spatial–spectral feature extraction network with patch attention module for hyperspectral image classification,

    R. Ji, K. Tan, X. Wang, C. Pan, and L. Xin, “Passnet: A spatial–spectral feature extraction network with patch attention module for hyperspectral image classification,” IEEE Geoscience and Remote Sensing Letters , vol. 20, pp. 1–5, 2023

    2023

  39. [50]

    RX anomaly detector with rectified background,

    M. Imani, “RX anomaly detector with rectified background,” IEEE Geoscience and Remote Sensing Letters , vol. 14, no. 8, pp. 1313–1317, 2017

    2017

  40. [51]

    Kernel RX-algorithm: A nonlinear anomaly detector for hyperspectral imagery,

    H. Kwon and N. M. Nasrabadi, “Kernel RX-algorithm: A nonlinear anomaly detector for hyperspectral imagery,” IEEE transactions on Geoscience and Remote Sensing , vol. 43, no. 2, pp. 388–397, 2005

    2005

  41. [52]

    A locally adaptive background density estimator: An evolution for RX-based anomaly detectors,

    S. Matteoli, T. Veracini, M. Diani, and G. Corsini, “A locally adaptive background density estimator: An evolution for RX-based anomaly detectors,” IEEE Geoscience and Remote Sensing Letters , vol. 11, no. 1, pp. 323–327, 2014

    2014

  42. [53]

    Fractional fourier transform- based tensor RX for hyperspectral anomaly detection,

    L. Zhang, J. Ma, B. Cheng, and F. Lin, “Fractional fourier transform- based tensor RX for hyperspectral anomaly detection,” Remote Sensing, vol. 14, no. 3, p. 797, 2022

    2022

  43. [54]

    Recursive RX with extended multi-attribute profiles for hyperspectral anomaly detection,

    F. He, S. Yan, Y . Ding, Z. Sun, J. Zhao, H. Hu, and Y . Zhu, “Recursive RX with extended multi-attribute profiles for hyperspectral anomaly detection,” Remote Sensing, vol. 15, no. 3, 2023

    2023

  44. [55]

    Adaptive reference-related graph embedding for hyperspectral anomaly detection,

    Y . Ma, S. Cai, and J. Zhou, “Adaptive reference-related graph embedding for hyperspectral anomaly detection,” IEEE Transactions on Geoscience and Remote Sensing , vol. 61, pp. 1–14, 2023

    2023

  45. [56]

    Collaborative representation for hyperspectral anomaly detection,

    W. Li and Q. Du, “Collaborative representation for hyperspectral anomaly detection,” IEEE Transactions on Geoscience and Remote Sensing, vol. 53, no. 3, pp. 1463–1474, 2015

    2015

  46. [57]

    Anomaly detection in hyperspectral images based on low-rank and sparse representation,

    Y . Xu, Z. Wu, J. Li, A. Plaza, and Z. Wei, “Anomaly detection in hyperspectral images based on low-rank and sparse representation,” IEEE Transactions on Geoscience and Remote Sensing , vol. 54, no. 4, pp. 1990–2000, 2016

    1990

  47. [58]

    Effective anomaly space for hyperspectral anomaly detec- tion,

    C.-I. Chang, “Effective anomaly space for hyperspectral anomaly detec- tion,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–24, 2022

    2022

  48. [59]

    Hyperspectral anomaly detection based on chessboard topology,

    L. Gao, X. Sun, X. Sun, L. Zhuang, Q. Du, and B. Zhang, “Hyperspectral anomaly detection based on chessboard topology,” IEEE Transactions on Geoscience and Remote Sensing , vol. 61, pp. 1–16, 2023

    2023

  49. [60]

    Learning tensor low-rank representation for hyperspectral anomaly detection,

    M. Wang, Q. Wang, D. Hong, S. K. Roy, and J. Chanussot, “Learning tensor low-rank representation for hyperspectral anomaly detection,” IEEE Transactions on Cybernetics , vol. 53, no. 1, pp. 679–691, 2023

    2023

  50. [61]

    Enhanced total variation reg- ularized representation model with endmember background dictionary for hyperspectral anomaly detection,

    C. Zhao, C. Li, S. Feng, and X. Jia, “Enhanced total variation reg- ularized representation model with endmember background dictionary for hyperspectral anomaly detection,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–12, 2022

    2022

  51. [62]

    FusAtNet: Dual attention based spectrospatial multimodal fusion network for hyperspec- tral and lidar classification,

    S. Mohla, S. Pande, B. Banerjee, and S. Chaudhuri, “FusAtNet: Dual attention based spectrospatial multimodal fusion network for hyperspec- tral and lidar classification,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops , Jun...

    2020

  52. [63]

    Enhanced autoencoders with attention-embedded degradation learning for unsupervised hyperspectral image super-resolution,

    L. Gao, J. Li, K. Zheng, and X. Jia, “Enhanced autoencoders with attention-embedded degradation learning for unsupervised hyperspectral image super-resolution,” IEEE Transactions on Geoscience and Remote Sensing, vol. 61, pp. 1–17, 2023

    2023

  53. [64]

    Background-guided deformable convolutional autoencoder for hyperspectral anomaly detection,

    Z. Wu, M. E. Paoletti, H. Su, X. Tao, L. Han, J. M. Haut, and A. Plaza, “Background-guided deformable convolutional autoencoder for hyperspectral anomaly detection,” IEEE Transactions on Geoscience and Remote Sensing , vol. 61, pp. 1–16, 2023

    2023

  54. [65]

    Auto-AD: Autonomous hyperspectral anomaly detection network based on fully convolutional autoencoder,

    S. Wang, X. Wang, L. Zhang, and Y . Zhong, “Auto-AD: Autonomous hyperspectral anomaly detection network based on fully convolutional autoencoder,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–14, 2022

    2022

  55. [66]

    GAN-based hyperspectral anomaly detection,

    S. Arisoy, N. M. Nasrabadi, and K. Kayabol, “GAN-based hyperspectral anomaly detection,” in 2020 28th European Signal Processing Confer- ence (EUSIPCO), 2021, pp. 1891–1895

    2020

  56. [67]

    Dual- channel capsule generation adversarial network for hyperspectral image classification,

    J. Wang, S. Guo, R. Huang, L. Li, X. Zhang, and L. Jiao, “Dual- channel capsule generation adversarial network for hyperspectral image classification,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–16, 2021

    2021

  57. [68]

    Semisupervised spectral learning with generative adversarial network for hyperspectral anomaly detection,

    K. Jiang, W. Xie, Y . Li, J. Lei, G. He, and Q. Du, “Semisupervised spectral learning with generative adversarial network for hyperspectral anomaly detection,” IEEE Transactions on Geoscience and Remote Sensing, vol. 58, no. 7, pp. 5224–5236, 2020

    2020

  58. [69]

    Convolutional transformer- inspired autoencoder for hyperspectral anomaly detection,

    Z. He, D. He, M. Xiao, A. Lou, and G. Lai, “Convolutional transformer- inspired autoencoder for hyperspectral anomaly detection,” IEEE Geo- science and Remote Sensing Letters , vol. 20, pp. 1–5, 2023

    2023

  59. [70]

    Hyperspectral anomaly detection based on variational background inference and generative adversarial network,

    Z. Wang, X. Wang, K. Tan, B. Han, J. Ding, and Z. Liu, “Hyperspectral anomaly detection based on variational background inference and generative adversarial network,” Pattern Recognition, vol. 143, p. 109795, 2023. [Online]. Available: https://www.sciencedirect. com/science/ar...

    2023

  60. [71]

    Variational continual learning,

    C. V . Nguyen, Y . Li, T. D. Bui, and R. E. Turner, “Variational continual learning,” arXiv preprint arXiv:1710.10628 , 2017

    2017 arXiv

  61. [72]

    Discrimination among semi-arid landscape endmembers using the spectral angle mapper (SAM) algorithm,

    R. H. Yuhas, A. F. Goetz, and J. W. Boardman, “Discrimination among semi-arid landscape endmembers using the spectral angle mapper (SAM) algorithm,” in JPL, Summaries of the Third Annual JPL Airborne Geoscience Workshop. Volume 1: AVIRIS Workshop, 1992

    1992

  62. [73]

    Autoencoders, minimum description length and helmholtz free energy,

    G. E. Hinton and R. Zemel, “Autoencoders, minimum description length and helmholtz free energy,” Advances in neural information processing systems, vol. 6, 1993

    1993

  63. [74]

    Tanet: An unsupervised two-stream autoencoder network for hyperspectral unmixing,

    Q. Jin, Y . Ma, X. Mei, and J. Ma, “Tanet: An unsupervised two-stream autoencoder network for hyperspectral unmixing,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–15, 2021

    2021

  64. [75]

    Target detection with unconstrained linear mixture model and hierarchical denoising autoencoder in hyperspectral imagery,

    Y . Li, Y . Shi, K. Wang, B. Xi, J. Li, and P. Gamba, “Target detection with unconstrained linear mixture model and hierarchical denoising autoencoder in hyperspectral imagery,” IEEE Transactions on Image Processing, vol. 31, pp. 1418–1432, 2022

    2022

  65. [76]

    Maximum likelihood from incomplete data via the em algorithm,

    A. P. Dempster, N. M. Laird, and D. B. Rubin, “Maximum likelihood from incomplete data via the em algorithm,” Journal of the royal statistical society: series B (methodological) , vol. 39, no. 1, pp. 1–22, 1977

    1977

  66. [77]

    Hyperspectral anomaly detection with attribute and edge-preserving filters,

    X. Kang, X. Zhang, S. Li, K. Li, J. Li, and J. A. Benediktsson, “Hyperspectral anomaly detection with attribute and edge-preserving filters,” IEEE Transactions on Geoscience and Remote Sensing , vol. 55, no. 10, pp. 5600–5611, 2017

    2017

  67. [78]

    You only train once: Learning a general anomaly enhancement network with random masks for hyperspectral anomaly detection,

    Z. Li, Y . Wang, C. Xiao, Q. Ling, Z. Lin, and W. An, “You only train once: Learning a general anomaly enhancement network with random masks for hyperspectral anomaly detection,” IEEE Transactions on Geoscience and Remote Sensing , vol. 61, pp. 1–18, 2023

    2023

  68. [79]

    Memory aware synapses: Learning what (not) to forget,

    R. Aljundi, F. Babiloni, M. Elhoseiny, M. Rohrbach, and T. Tuytelaars, “Memory aware synapses: Learning what (not) to forget,” inProceedings of the European Conference on Computer Vision (ECCV) , September 2018, pp. 139–154

    2018

  69. [80]

    Overcoming catastrophic forgetting in neural networks,

    J. Kirkpatrick, R. Pascanu, N. Rabinowitz, J. Veness, G. Desjardins, A. A. Rusu, K. Milan, J. Quan, T. Ramalho, A. Grabska-Barwinska, D. Hassabis, C. Clopath, D. Kumaran, and R. Hadsell, “Overcoming catastrophic forgetting in neural networks,” Proceedings of the National Acade...

    2017

  70. [81]

    Continual learning of context- dependent processing in neural networks,

    G. Zeng, Y . Chen, B. Cui, and S. Yu, “Continual learning of context- dependent processing in neural networks,” Nature Machine Intelligence, vol. 1, no. 8, pp. 364–372, 2019

    2019

  71. [82]

    CL-CaGAN: Capsule differential adversarial continual learning for cross-domain hyperspectral anomaly detection,

    J. Wang, S. Guo, Z. Hua, R. Huang, J. Hu, and M. Gong, “CL-CaGAN: Capsule differential adversarial continual learning for cross-domain hyperspectral anomaly detection,” IEEE Transactions on Geoscience and Remote Sensing, vol. 62, pp. 1–15, 2024. JOURNAL OF LATEX CLASS FILES, V...

    2024

  72. [83]

    Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine,

    M. H. Zweig and G. Campbell, “Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine,” Clinical chemistry, vol. 39, no. 4, pp. 561–577, 1993

    1993

  73. [84]

    A coherent interpretation of auc as a measure of aggregated classification performance,

    C. Ferri, J. Hern ´andez-Orallo, and P. A. Flach, “A coherent interpretation of auc as a measure of aggregated classification performance,” in Proceedings of the 28th International Conference on Machine Learning (ICML-11), 2011, pp. 657–664

    2011

  74. [85]

    Component decomposition analysis for hyperspectral anomaly detection,

    S. Chen, C.-I. Chang, and X. Li, “Component decomposition analysis for hyperspectral anomaly detection,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–22, 2021. - Jianing Wang (Member, IEEE) received the B.S. and M.S. degrees in circuit and system from La...

    2021

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.