pith. sign in

arxiv: 2401.04139 · v4 · submitted 2024-01-07 · 💻 cs.LG

CCNETS: A Modular Causal Learning Framework for Pattern Recognition in Imbalanced Datasets

Hanbeot Park , Yunjeong Cho , Hunhee Kim This is my paper

Pith reviewed 2026-05-24 04:29 UTC · model grok-4.3

classification 💻 cs.LG
keywords causal cooperative networksimbalanced datasetsdata synthesispattern recognitionfeedback loopfraud detectionpredictive maintenanceZoint mechanism
0
0 comments X

The pith

CCNETS links classification outcomes to data synthesis through a causal feedback loop to improve rare-class detection in imbalanced settings.

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

The paper introduces CCNETS to solve the distribution mismatch that arises when generative models operate separately from classifiers in imbalanced pattern recognition tasks. It builds a three-module system—Explainer for latent abstraction, Reasoner for label prediction, and Producer for synthesis—that interacts through a dynamic causal feedback loop. Classification results directly shape the next round of sample generation to strengthen weak decision boundaries. A Zoint mechanism fuses latent and observable features during this process. On two real datasets with extreme imbalance, the framework reports higher F1-scores and AUPRC than standard baselines while also showing better generalization when training data is scarce.

Core claim

CCNETS establishes a functional causal link between generation, inference, and reconstruction by composing an Explainer, a Reasoner, and a Producer that interact through a dynamic causal feedback loop in which classification outcomes directly guide targeted sample synthesis; the Zoint mechanism performs adaptive fusion of latent and observable features, and this arrangement produces higher F1-scores and AUPRC than decoupled baselines on the Credit Card Fraud Detection dataset (fraud rate < 0.2 %) and the AI4I 2020 Predictive Maintenance dataset (failure rate < 4 %).

What carries the argument

The dynamic causal feedback loop in which classification outcomes directly guide targeted sample synthesis to adaptively reinforce vulnerable decision boundaries, together with the Zoint mechanism for adaptive fusion of latent and observable features.

If this is right

  • CCNETS produces higher F1-scores and AUPRC than baseline methods on the Credit Card Fraud Detection dataset.
  • CCNETS produces higher F1-scores and AUPRC than baseline methods on the AI4I 2020 Predictive Maintenance dataset.
  • Data synthesized by CCNETS yields improved generalization and learning stability under limited-data conditions.
  • The framework aligns synthetic data generation with classifier objectives in a scalable and interpretable manner.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The modular separation of Explainer, Reasoner, and Producer could allow individual components to be swapped for domain-specific architectures without retraining the entire system.
  • If the feedback loop remains stable, the approach might reduce dependence on hand-crafted oversampling rules such as SMOTE in other imbalanced domains.
  • Testing the same causal structure on medical or cybersecurity datasets with comparable rarity rates would reveal whether the alignment benefit transfers beyond the two reported tasks.

Load-bearing premise

The dynamic causal feedback loop will adaptively reinforce decision boundaries without introducing bias or instability.

What would settle it

Run the full CCNETS against an ablated version that removes the causal link between classification and synthesis on the Credit Card Fraud Detection dataset; if the ablated version shows no drop in F1-score or AUPRC, the claimed benefit of the loop is falsified.

read the original abstract

Handling class imbalance remains a central challenge in machine learning, particularly in pattern recognition tasks where identifying rare but critical anomalies is of paramount importance. Traditional generative models often decouple data synthesis from classification,leading to a distribution mismatch that limits their practical benefit. To address these shortcomings, we introduce Causal Cooperative Networks (CCNETS), a modular framework that establishes a functional causal link between generation, inference, and reconstruction. CCNETS is composed of three specialized cooperative modules: an Explainer for latent feature abstraction, a Reasoner for probabilistic label prediction, and a Producer for context-aware data synthesis. These components interact through a dynamic causal feedback loop, where classification outcomes directly guide targeted sample synthesis to adaptively reinforce vulnerable decision boundaries. A key innovation, our proposed Zoint mechanism, enables the adaptive fusion of latent and observable features, enhancing semantic richness and decisionmaking robustness under uncertainty. We evaluated CCNETS on two distinct real-world datasets: Credit Card Fraud Detection dataset, characterized by extreme imbalance (fraud rate < 0.2%), and the AI4I 2020 Predictive Maintenance dataset (failure rate < 4%). Across comprehensive experimental setups, CCNETS consistently outperformed baseline methods, achieving superior F1-scores, and AUPRC. Furthermore, data synthesized by CCNETS demonstrated enhanced generalization and learning stability under limited data conditions. These results establish CCNETS as a scalable, interpretable, and hybrid soft computing framework that effectively aligns synthetic data with classifier objectives, advancing robust imbalanced learning.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 3 minor

Summary. The paper introduces Causal Cooperative Networks (CCNETS), a modular framework with three cooperative modules—an Explainer for latent feature abstraction, a Reasoner for probabilistic label prediction, and a Producer for context-aware data synthesis—that interact via a dynamic causal feedback loop in which classification outcomes guide targeted sample synthesis. A Zoint mechanism is proposed for adaptive fusion of latent and observable features. The central claim is that this causal alignment yields superior F1-scores and AUPRC over baselines on the Credit Card Fraud Detection dataset (fraud rate <0.2%) and the AI4I 2020 Predictive Maintenance dataset (failure rate <4%), with synthesized data also showing improved generalization under limited-data conditions.

Significance. If the empirical superiority and loop stability are demonstrated with proper controls, the work could contribute to imbalanced learning by providing a hybrid soft-computing approach that couples generation and inference causally rather than decoupling them. The modular design offers potential interpretability advantages, though the manuscript supplies no parameter-free derivations, machine-checked proofs, or reproducible code to strengthen this assessment.

major comments (2)
  1. [Abstract] Abstract: The headline claim that 'CCNETS consistently outperformed baseline methods, achieving superior F1-scores, and AUPRC' is unsupported by any experimental details, baseline descriptions, statistical tests, ablation studies, or implementation specifics for the causal loop. This is load-bearing for the central empirical claim.
  2. [Abstract] Abstract: No convergence analysis, regularization terms, or stability guarantees are supplied for the dynamic causal feedback loop, despite the acknowledged risk of error amplification when guiding synthesis on extreme imbalance (<0.2% fraud rate). The weakest assumption—that the loop reinforces boundaries without introducing bias or instability—therefore remains unexamined.
minor comments (3)
  1. [Abstract] Abstract: Missing space after comma in 'classification,leading to a distribution mismatch'.
  2. [Abstract] Abstract: 'decisionmaking' should be hyphenated as 'decision-making'.
  3. [Abstract] Abstract: The phrase 'comprehensive experimental setups' is used without elaboration on what those setups entail.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address the two major comments on the abstract below, indicating planned revisions to strengthen the presentation of claims and the causal loop.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The headline claim that 'CCNETS consistently outperformed baseline methods, achieving superior F1-scores, and AUPRC' is unsupported by any experimental details, baseline descriptions, statistical tests, ablation studies, or implementation specifics for the causal loop. This is load-bearing for the central empirical claim.

    Authors: The abstract is a concise summary; the full manuscript provides the requested details in the Experiments section, including baseline methods, F1-scores, AUPRC values, statistical comparisons, ablation studies, and implementation specifics for the causal loop. To address the concern that the claim appears unsupported within the abstract itself, we will revise the abstract to incorporate brief quantitative highlights and references to the supporting experimental evidence. revision: yes

  2. Referee: [Abstract] Abstract: No convergence analysis, regularization terms, or stability guarantees are supplied for the dynamic causal feedback loop, despite the acknowledged risk of error amplification when guiding synthesis on extreme imbalance (<0.2% fraud rate). The weakest assumption—that the loop reinforces boundaries without introducing bias or instability—therefore remains unexamined.

    Authors: The manuscript does not supply a formal convergence analysis, explicit regularization terms, or theoretical stability guarantees for the dynamic causal feedback loop. Empirical results on the two highly imbalanced datasets are presented as evidence of practical performance. We will revise the manuscript by adding a discussion subsection on observed loop behavior, any regularization used in training, and an explicit acknowledgment of the lack of theoretical guarantees, noting this as a limitation and direction for future work. revision: yes

Circularity Check

0 steps flagged

No significant circularity; empirical framework with independent experimental validation

full rationale

The paper proposes CCNETS as a new modular architecture (Explainer-Reasoner-Producer with Zoint and feedback loop) and reports its performance on two fixed external datasets. No first-principles derivation, mathematical prediction, or uniqueness theorem is claimed that reduces to the inputs by construction. The design choices are presented as innovations, and the superiority claims rest on experimental outcomes that are not forced by the architecture definition itself; the results could falsify the method. No self-citation chains or fitted-input renamings appear in the load-bearing steps.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 4 invented entities

The central claim rests on the unproven domain assumption that causal feedback between classification and synthesis will improve boundary reinforcement, plus multiple invented components whose value is asserted without independent evidence beyond the reported experiments.

free parameters (1)
  • Module hyperparameters and fusion weights
    The framework description implies numerous tunable parameters in the Explainer, Reasoner, Producer, and Zoint mechanism that must be fitted to achieve the reported performance.
axioms (1)
  • domain assumption A functional causal link between generation, inference, and reconstruction improves alignment of synthetic data with classifier objectives on imbalanced tasks.
    This premise is invoked to justify the dynamic feedback loop design and is not derived within the abstract.
invented entities (4)
  • Explainer module no independent evidence
    purpose: Latent feature abstraction
    New specialized component introduced as part of the modular framework.
  • Reasoner module no independent evidence
    purpose: Probabilistic label prediction
    New specialized component introduced as part of the modular framework.
  • Producer module no independent evidence
    purpose: Context-aware data synthesis
    New specialized component introduced as part of the modular framework.
  • Zoint mechanism no independent evidence
    purpose: Adaptive fusion of latent and observable features
    Key innovation presented as enabling semantic richness under uncertainty.

pith-pipeline@v0.9.0 · 5808 in / 1668 out tokens · 28971 ms · 2026-05-24T04:29:44.850023+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

45 extracted references · 45 canonical work pages · 3 internal anchors

  1. [1]

    Turhan, H.S

    C.G. Turhan, H.S. Bilge, Recent Trends in Deep Generative Models: a Review, in: 2018 3rd International Conference on Computer Science and Engineering (UBMK), IEEE, 2018: pp. 574–579. https://doi.org/10.1109/UBMK.2018.8566353

    work page doi:10.1109/ubmk.2018.8566353 2018

  2. [2]

    Z. Xing, O. Mehmood, W.A.P. Smith, Unsupervised anomaly detection with a temporal continuation, confidenc e-aware V AE -GAN, Pattern Recognit 166 (2025). https://doi.org/10.1016/j.patcog.2025.111699

    work page doi:10.1016/j.patcog.2025.111699 2025

  3. [3]

    Qiang, J

    N. Qiang, J. Gao, Q. Dong, H. Yue, H. Liang, L. Liu, J. Yu, J. Hu, S. Zhang, B. Ge, Y . Sun, Z. Liu, T. Liu, J. Li, H. Song, S. Zhao, Functional brain network identification and fMRI augmentation using a V AE -GAN framework, Comput Biol Med 165 (2023). https://doi.org/10.1016/j.compbiomed.2023.107395

    work page doi:10.1016/j.compbiomed.2023.107395 2023

  4. [4]

    Y . Han, Y . Liu, Q. Chen, Data augmentation in material images using the improved HP- V AE-GAN, Comput Mater Sci 22 6 (2023). https://doi.org/10.1016/j.commatsci.2023.112250

    work page doi:10.1016/j.commatsci.2023.112250 2023

  5. [5]

    Goodfellow, J

    I.J. Goodfellow, J. Pouget -Abadie, M. Mirza, B. Xu, D. Warde -Farley, S. Ozair, A. Courville, Y . Bengio, Generative Adversarial Nets, in: Advances in Neural Information Processing Systems 27 (NI PS 2014), 2014: pp. 1–9. http://www.github.com/goodfeli/adversarial

    work page 2014

  6. [6]

    Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

    A. Radford, L. Metz, S. Chintala, Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks, (2015). https://doi.org/10.48550/arXiv.1511.06434

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1511.06434 2015

  7. [7]

    Lauer and C

    T.R. Shaham, T. Dekel, T. Michaeli, SinGAN: Learning a Generative Model From a Single Natural Image, in: 2019 IEEE/CVF International Conference on Computer Vision (ICCV), IEEE, 2019: pp. 4569–4579. https://doi.org/10.1109/ICCV .2019.00467

    work page doi:10.1109/iccv 2019

  8. [8]

    A Style-Based Generator Architecture for Generative Adversarial Networks

    T. Karras, S. Laine, T. Aila, A Style -Based Generator Architecture for Generative Adversarial Networks, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2019. (2019) 4401–4410. https://doi.org/10.48550/arXiv.1812.04948

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1812.04948 2019

  9. [9]

    Sasipriyaa, P

    N. Sasipriyaa, P. Natesan, E. Gothai, G. Madhesan, E. Madhumitha, K. V . Mithun, Recognition of Tami l handwritten characters using Scrabble GAN, in: 2023 International Conference on Computer Communication and Informatics, ICCCI 2023, Institute of Electrical and Electronics Engineers Inc., 2023. https://doi.org/10.1109/ICCCI56745.2023.10128564

    work page doi:10.1109/iccci56745.2023.10128564 2023

  10. [10]

    Kowsari, S.J

    Y . Kowsari, S.J. Mahdavi Chabok, M.H. Moattar, Classification of Pulmonary Images By Using Generative Adversarial Networks, in: 2020 8th Iranian Joint Congress on Fuzzy and Intelligent Systems (CFIS), IEEE, 2020: pp. 133–137. https://doi.org/10.1109/CFIS49607.2020.9238755

    work page doi:10.1109/cfis49607.2020.9238755 2020

  11. [11]

    Sampath Kumar, V .P

    R. Sampath Kumar, V .P. Krishnamurthy, V . Podile, G. Yamini Priyanka, V . Neha, Generative Adversarial Networks to Improve the Nature of Training in Autonomous Vehicles, in: 2023 International Conference on Disruptive Technologies, ICDT 202 3, Institute of Electrical and Electronics Engineers Inc., 2023: pp. 161–164. https://doi.org/10.1109/ICDT57929.202...

    work page doi:10.1109/icdt57929.2023.10151288 2023

  12. [12]

    Zheng, C

    S. Zheng, C. Gupta, Discriminant Generative Adversarial Networks with its Application to Equipment Health Classification, in: ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), IEEE, 2020: pp. 3067–3071. https://doi.org/10.1109/ICASSP40776.2020.9053475

    work page doi:10.1109/icassp40776.2020.9053475 2020

  13. [13]

    Y . Shi, Q. Li, X.X. Zhu, Building Footprint Generation Using Improve d Generative Adversarial Networks, IEEE Geoscience and Remote Sensing Letters. 16 (2019) 603–

    work page 2019

  14. [14]

    https://doi.org/10.1109/LGRS.2018.2878486

    work page doi:10.1109/lgrs.2018.2878486 2018

  15. [15]

    X. Chen, R. Roberts, Z. Liu, W. Tong, A generative adversarial network model alternative to animal studies for clinical pathology assessment, Nat Commun. 14 (2023). https://doi.org/10.1038/s41467-023-42933-9

    work page doi:10.1038/s41467-023-42933-9 2023

  16. [16]

    Zhang, J

    Q. Zhang, J. Yang, X. Zhang, T. Cao, Generating Adversarial Examples in Audio Classification with Generative Adversarial Network, in: 2022 7th Internati onal Conference on Image, Vision and Computing, ICIVC 2022, Institute of Electrical and Electronics Engineers Inc., 2022: pp. 848–853. https://doi.org/10.1109/ICIVC55077.2022.9886154

    work page doi:10.1109/icivc55077.2022.9886154 2022

  17. [17]

    Shastri, C

    P. Shastri, C. Patil, P. Wanere, S. Mahajan, A. Bhatt, Adversarial Synthesis based Data Augmentation for Speech Classification, in: 2022 International Conference on Signal and Information Processing, IConSIP 2022, Institute of Electrical and Electronics Engineers Inc., 2022. https://doi.org/10.1109/ICoNSIP49665.2022.10007491

    work page doi:10.1109/iconsip49665.2022.10007491 2022

  18. [18]

    Y . Zhan, Y . Wang, X. Yu, Semisupervised hyperspectral image classification based on generative adversarial networks and spectral angle distance, Sci Rep. 13 (2023). https://doi.org/10.1038/s41598-023-49239-2

    work page doi:10.1038/s41598-023-49239-2 2023

  19. [19]

    Z. Yu, W. Cui, Robust hyperspectral image classification using generative adversarial networks, Inf Sci (N Y) 666 (2024). https://doi.org/10.1016/j.ins.2024.120452

    work page doi:10.1016/j.ins.2024.120452 2024

  20. [20]

    H. Gao, H. Zhang, X. Yang, W. Li, F. Gao, Q. Wen, Generatin g natural adversarial examples with universal perturbations for text classification, Neurocomputing. 471 (2022) 175–182. https://doi.org/10.1016/j.neucom.2021.10.089

    work page doi:10.1016/j.neucom.2021.10.089 2022

  21. [21]

    Benítez, A.S

    R.G. Benítez, A.S. Navarrete, C. Vidal -Castro, C. Martínez -Araneda, Guide for the application of the data augmentation approach on sets of texts in Spanish for sentiment and emotion analysis, PLoS One 19 (2024). https://doi.org/10.1371/journal.pone.0310707

    work page doi:10.1371/journal.pone.0310707 2024

  22. [22]

    H.S. Choi, D. Jung, S. Kim, S. Yoon, Imbalanced Data Classification via Cooperative Interaction Between Classifier and Generator, IEEE Trans Neural Netw Learn Syst. 33 (2022) 3343–3356. https://doi.org/10.1109/TNNLS.2021.3052243

    work page doi:10.1109/tnnls.2021.3052243 2022

  23. [23]

    Wickramar atne, M.S

    S.D. Wickramar atne, M.S. Mahmud, LSTM based GAN Networks for Enhancing Ternary Task Classification Using fNIRS Data, in: 2021 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), IEEE, 2021: pp. 1043–1046. https://doi.org/10.1109/EMBC46164.2021.9630000

    work page doi:10.1109/embc46164.2021.9630000 2021

  24. [24]

    A.N. Dar, R. Rastogi, MLGAN: Addressing Imbalance in Multilabel Learning Using Generative Adversarial Networks, in: 2023 International Conference on Emerging Techniques in Computational Intelligence (ICETCI), IEEE, 2023: pp. 324–331. https://doi.org/10.1109/ICETCI58599.2023.10331105

    work page doi:10.1109/icetci58599.2023.10331105 2023

  25. [25]

    Lu, K.-L

    Y.-W. Lu, K.-L. Liu, C.-Y . Hsu, Conditional Generative Adversarial Network for Defect Classification with Class Imbalance, in: 2019 IEEE International Conference on Smart Manufacturing, Industrial & Logistics Engineering (SMILE), IEEE, 2019: pp. 146–149. https://doi.org/10.1109/SMILE45626.2019.8965320

    work page doi:10.1109/smile45626.2019.8965320 2019

  26. [26]

    C. Liu, X. Wang, K. Wu, J. Tan, F. Li, W. Liu, Oversampling for Imbalanced Time Series Classification Based on Generative Adversarial Networks, in: 2018 IEEE 4th International Conference on Computer and Communications (ICCC), IEEE, 2018: pp. 1104–1108. https://doi.org/10.1109/CompComm.2018.8780808

    work page doi:10.1109/compcomm.2018.8780808 2018

  27. [27]

    Neagoe, P

    V.-E. Neagoe, P. Diaconescu, CNN Hyperspectral Image Classification Using Training Sample Augmentation with Generative Adversarial Networks, in: 2020 13th International Conference on Communications (COMM), IEEE, 2020: pp. 515–519. https://doi.org/10.1109/COMM48946.2020.9142021

    work page doi:10.1109/comm48946.2020.9142021 2020

  28. [28]

    X. Zhu, Y . Liu, J. Li, T. Wan, Z. Qin, Emotion classification with data augmentation using generative adversarial networks, in: Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Springer Verlag, 2018: pp. 349–360. https://doi.org/10.1007/978-3- 319-93040-4_28

    work page doi:10.1007/978-3- 2018

  29. [29]

    Madhu, S

    A. Madhu, S. Kumaraswamy, Data Augmentation Using Generative Adversarial Network for Environmental Sound Classification, in: 2019 27th European Signal Processing Conference (EUSIPCO), IEEE, 2019: pp. 1–5. https://doi.org/10.23919/EUSIPCO.2019.8902819

    work page doi:10.23919/eusipco.2019.8902819 2019

  30. [30]

    Seeliger, U

    K. Seeliger, U. Güçlü, L. Ambrogioni, Y . Güçlütürk, M.A.J. van Gerven, Generative adversarial networks for reconstructing natural images from brain activity, Neuroimage. 181 (2018) 775–785. https://doi.org/10.1016/j.neuroimage.2018.07.043

    work page doi:10.1016/j.neuroimage.2018.07.043 2018

  31. [31]

    Gaziv, R

    G. Gaziv, R. Beliy, N. Granot, A. Hoogi, F. Strappini, T. Golan, M. Irani, Self-supervised Natural Image Reconstruction and Large -scale Semantic Classification from Brain Activity, Neuroimage. 254 (2022). https://doi.org/10.1016/j.neuroimage.2022.119121

    work page doi:10.1016/j.neuroimage.2022.119121 2022

  32. [32]

    Miyawaki, H

    Y . Miyawaki, H. Uchida, O. Yamashita, M. aki Sato, Y . Morito, H.C. Tanabe, N. Sadato, Y . Kamitani, Visual Image Reconstruction from Human Brain Activity using a Combination of Multiscale Local Image Decoders, Neuron. 60 (2008) 915–929. https://doi.org/10.1016/j.neuron.2008.11.004

    work page doi:10.1016/j.neuron.2008.11.004 2008

  33. [33]

    T. Dado, Y . Güçlütürk, L. Ambrogioni, G. Ras, S. Bosch, M. van Gerven, U. Güçlü, Hyperrealistic neural decoding for reconstructing faces from fMRI activations via the GAN latent space, Sci Rep. 12 (2022). https://doi.org/10.1038/s41598-021-03938-w

    work page doi:10.1038/s41598-021-03938-w 2022

  34. [34]

    Naselaris, R.J

    T. Naselaris, R.J. Prenger, K.N. Kay, M. Oliver, J.L. Gallant, Bayesian Reconstruction of Natural Images from Human Brain Activity, Neuron. 63 (2009) 902–915. https://doi.org/10.1016/j.neuron.2009.09.006

    work page doi:10.1016/j.neuron.2009.09.006 2009

  35. [35]

    Gopalakrishnan, P.R

    S. Gopalakrishnan, P.R. Singh, Y . Yazici, C.S. Foo, V . Chandrasekhar, A.M. Ambikapathi, Classify and generate: Using classification latent space representations for image generations, Neurocomputing. 471 (2022) 296–334. https://doi.org/10.1016/j.neucom.2021.10.090

    work page doi:10.1016/j.neucom.2021.10.090 2022

  36. [36]

    Nishimoto, A.T

    S. Nishimoto, A.T. Vu, T. Naselaris, Y . Benjamini, B. Yu, J.L. Gallant, Reconstructing visual experiences from brain activity evoked by natural movies, Current Biology. 21 (2011) 1641–1646. https://doi.org/10.1016/j.cub.2011.08.031

    work page doi:10.1016/j.cub.2011.08.031 2011

  37. [37]

    Schoenmakers, M

    S. Schoenmakers, M. Barth, T. Heskes, M. van Gerven, Linear reconstruction of perceived images from human brain activity, Neuroimage. 83 (2013) 951–961. https://doi.org/10.1016/j.neuroimage.2013.07.043

    work page doi:10.1016/j.neuroimage.2013.07.043 2013

  38. [38]

    C. Li, X. Wang, B. Miao, C. Xie, Z. Wang, Y . Zhu, An Efficient Framework for Enhancing Discriminative Models via Diffusion Techniques, Proceedings of the AAAI Conference on Artificial Intelligence 39 (2025) 4670–4678. https://doi.org/10.1609/aaai.v39i5.32493

    work page doi:10.1609/aaai.v39i5.32493 2025

  39. [39]

    Artiushenko, R

    V . Artiushenko, R. Reider, T. Reggelin, S. Lang, Advancing robotic systems with Ensemble and Modular Deep Learning: Research idea and framework, in: Procedia Comput Sci, Elsevier B.V ., 2025: pp. 2557–2566. https://doi.org/10.1016/j.procs.2025.01.315

    work page doi:10.1016/j.procs.2025.01.315 2025

  40. [40]

    S. Chen, H. Min, Y . Fang, X. Wu, B. Li, X. Zhao, Uncertainty -aware Sensor Data Anomaly Detection for Autonomous Vehicles, in: IEEE Intelligent Vehicles Symposium, Proceedings, Institute of Electrical and Electronics Engineers Inc., 2024: pp. 478–483. https://doi.org/10.1109/IV55156.2024.10588587

    work page doi:10.1109/iv55156.2024.10588587 2024

  41. [41]

    Ororbia, D

    A. Ororbia, D. Kifer, The neural coding framework for learning generative models, Nat Commun. 13 (2022). https://doi.org/10.1038/s41467-022-29632-7

    work page doi:10.1038/s41467-022-29632-7 2022

  42. [42]

    Bahmei, E

    B. Bahmei, E. Birmingham, S. Arzanpour, CNN -RNN and Data Augmentation Using Deep Convolutional Generative Adversarial Network for Environmental Sound Classification, IEEE Signal Process Lett 29 (2022) 682–686. https://doi.org/10.1109/LSP.2022.3150258

    work page doi:10.1109/lsp.2022.3150258 2022

  43. [43]

    H. Guo, R. Tang, Y . Ye, Z. Li, X. He, DeepFM: A Factorization-Machine based Neural Network for CTR Prediction, 2017. https://doi.org/10.48550/arXiv.1703.04247

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1703.04247 2017

  44. [44]

    Touvron, P

    H. Touvron, P. Bojanowski, M. Caron, M. Cord, A. El-Nouby, E. Grave, G. Izacard, A. Joulin, G. Synnaeve, J. Verbeek, H. Jegou, ResMLP: Feedforward Networks for Image Classification with Data -Efficient Training, IEEE Trans Pattern Anal Mach Intell. 45 (2023) 5314–5321. https://doi.org/10.1109/TPAMI.2022.3206148

    work page doi:10.1109/tpami.2022.3206148 2023

  45. [45]

    Y . Wang, L. Deng, Enhanced Data Augmentation for Infrared Images with Generative Adversarial Networks Aided by Pretrained Models, IEEE Access (2024). https://doi.org/10.1109/ACCESS.2024.3491167

    work page doi:10.1109/access.2024.3491167 2024