Pith. sign in

REVIEW 5 major objections 6 minor 54 references

Deep Feature Response Discriminative Calibration

T0 review · 5 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read Calibrating feature responses with Gaussian confidence values improves classification accuracy and feature distinctiveness over ResNet and SENet.

desk verdict A clearly described GELU/SENet variant whose headline accuracy claim is contradicted by its own Table 1; the GCLU ablation is a useful empirical sweep, but the comparisons need fixing before the central claim can be taken at face value. read the letter →

arxiv 2411.13582 v1 pith:2F2NUHRF submitted 2024-11-16 cs.CV

Wenxiang Xu , Tian Qiu , Linyun Zhou , Zunlei Feng , Mingli Song , Huiqiong Wang This is my paper · ORCID
classification cs.CV
keywords DeepNeuralNetworkResponseValueCalibrationGaussianDistributionFeatureDiscriminabilityResCNetActivationFunctionImageClassification
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

The paper proposes that a neural network's classification accuracy can be improved by calibrating each convolutional kernel's response values against a per-kernel Gaussian model. It claims that response values for a kernel are approximately Gaussian-distributed, and therefore a confidence weight can be computed for each response as a symmetric integral of the standard normal CDF around the learned mean. Multiplying the original response by this weight and adding the product back into the feature map yields a plugin 'Response Calibration Layer' that, inserted into ResNet blocks, produces ResCNet. The paper reports that ResCNet improves accuracy and feature discriminability over ResNet and SENet on CIFAR-10, CIFAR-100, SVHN, and ImageNet, and that with standard training tricks it exceeds 86% on CIFAR-100 with a ResNet-50 variant. A sympathetic reader would care because the method offers a distribution-based, parameter-lean alternative to scaling-based attention modules like SENet.

What carries the argument

The central object is the response calibration weight $w_k$ defined by the folded Gaussian CDF: for a feature response $a_k$ with per-kernel mean $\mu_k$ and standard deviation $\sigma_k$, let $x=(a_k-\mu_k)/\sigma_k$; then $w_k = \Phi(x)$ for $a_k\le\mu_k$ and $w_k = 1-\Phi(x)$ for $a_k>\mu_k$, where $\Phi$ is the standard normal CDF. This weight is close to 1 near the mean and falls toward 0 in the tails, so responses that are typical of the kernel are kept at nearly full strength while atypical responses are suppressed. The calibration value $c_k = a_k w_k$ is added to the original response, and a module containing a global-pooling step plus two or three fully-connected layers learns the per-kernel $\mu_k$ and $\sigma_k$. This module is inserted after the final BatchNorm in each residual block, mirroring the SE layer's position, but with addition rather than multiplicative scaling, giving the RC layer and the ResCNet architecture.

What would settle it

Retrain ResCNet-32 on CIFAR-100 with the learnable Gaussian CDF weight replaced by a fixed, symmetric, non-Gaussian weight of the same shape (for instance, a triangular function of the standardized value); if accuracy is unchanged, the specific Gaussian assumption is not the source of the improvement.

Watch

Extended reading notes

Core claim

The central claim is that each convolutional kernel's response values follow a Gaussian distribution $\mathcal{N}(\mu_k, \sigma_k^2)$, and that this law can be exploited by defining a confidence weight for each response $a_k$ as the Gaussian CDF evaluated at the standardized value $x=(a_k-\mu_k)/\sigma_k$, with the weight mirrored for values above the mean: $w_k = \Phi(x)$ if $a_k \le \mu_k$ and $w_k = 1-\Phi(x)$ otherwise. The calibration value is $c_k = a_k w_k$, and the RC layer adds this calibration value to the original feature (not just scales it, as SENet does). The paper develops this into a plugin module placed after the last BatchNorm in a residual block, with two or three fully-connected layers fitting $\mu_k$ and $\sigma_k$ per kernel, and calls the resulting architecture ResCNet. Across CIFAR-10/100, SVHN, and ImageNet, ResCNet is reported to beat both ResNet and SENet baselines in accuracy, and the paper claims stronger feature distinctiveness as shown by response-value distributions. On CIFAR-100 with ResNet-50 and a series of standard tricks, ResCNet reaches 86.31%, which the authors state is the first time a 50-layer ResNet variant has exceeded 86% without extra data or pre-training.

Load-bearing premise

The load-bearing premise is that each convolutional kernel's response values follow a Gaussian distribution; if that distributional assumption fails, the confidence values that drive the calibration are no longer meaningful and the module becomes an arbitrary nonlinearity.

Editorial extensions

If this is right

  • Because the RC layer is a plugin, the calibration mechanism can in principle be inserted into any convolutional backbone, not only ResNet.
  • The paper's activation-function variant, GCLU, applies the same symmetric CDF weighting as a standalone activation and is reported to improve accuracy on shallow CNNs like AlexNet that lack BatchNorm.
  • The reported 86.31% on CIFAR-100 with ResNet-50, if reproducible, marks a new performance point for a 50-layer ResNet variant without extra training data.
  • The method suggests that feature discriminability can be improved by distribution-based response correction rather than by the scaling-only approach of SENet.

Reading between the lines

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

  • The Gaussian assumption is asserted from a single violin plot and an appeal to the Central Limit Theorem; a natural extension would be to test other distributional families (e.g., logistic or heavy-tailed) for the same folded-weight construction, since the method's rationale would change if the true distributions deviate markedly from Gaussian.
  • Because the calibration values are added to the original features rather than multiplied, the RC layer can be viewed as injecting a learned 'typicality' signal; this is conceptually similar to a residual attention mechanism, and one could test whether the improvement comes from the distributional modeling or simply from the extra non-linearity and parameters.
  • The method could be probed on tasks beyond classification, such as object detection or segmentation, where feature distinctiveness matters, to see if the calibration transfers.
Share X Bluesky LinkedIn Reddit HN

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 Response Calibration Networks (ResCNet), a plug-in module that estimates a per-channel Gaussian model for convolutional feature responses, converts each response into a confidence weight via the standard normal CDF (with symmetric complementary weighting for above-mean responses), multiplies the original response by this weight, and adds the calibrated value to the original feature. The module is inserted after the second BatchNorm in a ResNet block, in a configuration analogous to but distinct from the SE layer. Experiments on CIFAR-10, CIFAR-100, SVHN, and ImageNet compare ResCNet with ResNet and SENet; a related activation-function variant, GCLU, is also evaluated across several architectures. The paper claims improved classification accuracy and improved feature discriminability.

Significance. If the central empirical comparison were clean, this would be a simple, plug-compatible feature-calibration module and a modestly interesting alternative to channel attention. The paper has concrete strengths: the code is publicly available, the experimental coverage spans four datasets, and the ablation study is fairly broad, including integration schemes, CDF approximations, and an activation-function variant. However, the current evidence does not establish the central claim. One of the key comparisons in the paper's own Table 1 contradicts the claim that ResCNet outperforms SENet; the ImageNet comparison is not epoch-matched; and no repeated-seed statistics or significance tests are reported. The conceptual novelty is also limited by the fact that the mean and standard deviation are learned parameters, so the calibrated transformation is a learned nonlinearity whose Gaussian interpretation is motivational unless quantitative distributional evidence is supplied.

major comments (5)
  1. [Section 5.2.1, Table 1] The central claim that 'as the depth of the model increases, the proposed ResCNet outperforms both ResNet and SENet' is contradicted by Table 1: on CIFAR-100, ResCNet-56 reaches 73.81%, which is 0.87 points below SENet-56 (74.68%) and only 0.02 points above ResNet-56 (73.79%). On CIFAR-10, ResNet-56 and SENet-56 both reach 94.51% while ResCNet-56 reaches 94.61%, a difference too small to count as an improvement in the absence of repeated runs. Please reconcile the stated claim with these numbers or explicitly narrow the scope of the claimed advantage.
  2. [Table 2 and Section 5.2.1] The ImageNet comparison is not controlled. SENet-50 is a published 100-epoch result, whereas ResCNet-50 is trained for 90 epochs; the ResNet-18 and ResNet-34 rows mix TorchVision pretrained checkpoints with in-house re-implementations; and parameter counts differ across methods. Without all baselines retrained under identical epochs, schedulers, augmentation, and hyperparameters, the statement that ResCNet-50 'outperformed SENet-50' is not justified. Please provide an epoch-matched comparison or both 90- and 100-epoch SENet baselines.
  3. [Section 5.2.1 and Tables 1-3] No standard deviations, multiple seeds, or significance tests are reported anywhere in the paper. Most reported gains are fractions of a percentage point, for example Table 1 shows CIFAR-10 ResCNet-32 at 93.79% versus SENet-32 at 93.73%, and Table 3 shows a baseline difference of 80.79% versus 80.87%. These differences are within typical run-to-run variation for CIFAR training. Please report at least three independent seeds with means and standard deviations, or provide a significance test, before claiming superiority over the baselines.
  4. [Section 3 and Eqs. (1)-(4)] The Gaussian assumption is supported only by visual inspection of a single violin plot and a CLT argument that does not account for dependencies, ReLU truncation, or the fact that the mean and standard deviation are learned parameters. Because Eq. (4) defines the calibration weights as Gaussian CDF integrals, this assumption is load-bearing for the stated rationale of the mechanism. Please provide a quantitative distributional test (for example, a goodness-of-fit check across multiple layers and datasets), or explicitly reframe the RC Layer as a learned nonlinearity whose Gaussian interpretation is only heuristic.
  5. [Section 5.2.2, Fig. 5] The claim that the calibrated features are 'more discriminative' is supported only by red dashed bounding boxes in a single visualization. Since discriminability is central to the paper's title and contribution, please report a quantitative metric, such as between-class versus within-class feature variance, k-nearest-neighbor classification accuracy on frozen features, or a margin-based measure, to substantiate the claim.
minor comments (6)
  1. [Section 5.3.1] The parameter-count argument for the three-FC-layer RC Layer states that 'the total parameter count becomes 3 * C * r'. For the described C -> C/r -> C architecture, the dominant term is 2*C^2/r, so the formula and the accompanying reasoning should be corrected and recomputed.
  2. [Table 1] The table header 'Params (K) FLOPs (M)' is shared between the 'CIFAR-10 & SVHN' and 'CIFAR-100' dataset columns in a confusing way. Please separate architecture statistics from dataset-specific classifier statistics and clarify how the FLOPs are computed.
  3. [Table 5] AlexNet with GELU on CIFAR-100 is reported at 1.23% accuracy, which strongly suggests a training failure. This row should either be discussed in the text or removed, because it distorts the activation-function comparison.
  4. [Section 5.2.3] The claim of being 'the first to achieve over 86% accuracy on the CIFAR-100 dataset with a 50-layer ResNet model variant' needs a verified, dated comparison with other published results rather than a leaderboard footnote, especially because the hyperparameters were selected on the same test set.
  5. [Section 5.1.2] Momentum, weight decay, and exact data-augmentation protocols for CIFAR and SVHN are not reported, although these choices materially affect the reported accuracy values. Please specify them or refer explicitly to the released code.
  6. [Throughout] There are numerous typographical and formatting issues, including 'Comparision' in Section 5.2.1, 'CIF AR' in place of 'CIFAR', 'T op-1' in Figure 4, and inconsistent capitalization of 'CaIT' versus 'Cait'. These should be corrected in a final revision.

Circularity Check

0 steps flagged · score 0.0 of 10

The derivation chain is self-contained: the calibration values are a learned nonlinear transform of each response, and the accuracy/discriminability claims are empirical measurements rather than predictions forced by the fitted parameters.

full rationale

No circular step reduces a claimed prediction to its own inputs. The paper's calibration mechanism (Eqs. 1–5) defines the weight w_k as a Gaussian CDF evaluated on the standardized response (a_k − µ_k)/σ_k, with µ_k and σ_k learned per kernel; adding a_k*w_k back to a_k is a trainable elementwise nonlinearity. The reported accuracies (Tables 1–3) are measured outcomes of training, not quantities derived from the fitted µ and σ, so there is no fitted-input-called-prediction structure. The Gaussian-response premise is asserted from a violin plot and CLT handwaving rather than rigorously established, but that is an unsupported assumption, not circular reasoning. The only author self-citation, Ref. [14], appears in a list of object-recognition applications and is not load-bearing for the method or its evaluation. The central claim is therefore independently testable: the empirical comparison with ResNet/SENet, while contestable on statistical grounds (single runs, non-epoch-matched ImageNet SENet row), does not reduce by construction.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The method rests on an untested Gaussian assumption for feature responses, learned per-channel distribution parameters, and an ad hoc symmetric weighting scheme. These are inputs to the method, not results derived from it.

free parameters (3)
  • Per-channel mean μ_k = learned during training
    Estimated by the MLP in the RC Layer for each convolutional kernel; used to center the Gaussian CDF.
  • Per-channel standard deviation σ_k = learned during training
    Estimated by the MLP; scales the argument of the Gaussian CDF.
  • Reduction ratio r for the RC Layer bottleneck = 4 for ResNet-32/56, 16 for ResNet-18/34/50
    Chosen by hand to trade off parameter count and capacity; affects the fitted mean/variance representation.
assumptions (3)
  • domain assumption Feature responses of each convolutional kernel follow a Gaussian distribution.
    Stated in Section 3 and used to define calibration weights in Eq. 1-5. No goodness-of-fit test is provided.
  • ad hoc to paper The symmetric CDF weighting for above-mean responses (1-Phi(x)) is the correct way to measure confidence.
    Introduced in Eq. 4 without derivation; it makes the weight decrease away from the mean in both tails but is one of many possible choices.
  • domain assumption The MLP in the RC Layer can accurately estimate the mean and variance of each kernel's response distribution.
    Relied on in Section 4.2; no analysis of estimation error is given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep Feature Response Discriminative Calibration." pith.science (2026). https://pith.science/paper/2F2NUHRF

@misc{pith2026241113582,
  author       = {Pith},
  title        = {Pith review of: Deep Feature Response Discriminative Calibration},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2F2NUHRF}},
  note         = {Machine review of arXiv:2411.13582}
}
read the original abstract

Deep neural networks (DNNs) have numerous applications across various domains. Several optimization techniques, such as ResNet and SENet, have been proposed to improve model accuracy. These techniques improve the model performance by adjusting or calibrating feature responses according to a uniform standard. However, they lack the discriminative calibration for different features, thereby introducing limitations in the model output. Therefore, we propose a method that discriminatively calibrates feature responses. The preliminary experimental results indicate that the neural feature response follows a Gaussian distribution. Consequently, we compute confidence values by employing the Gaussian probability density function, and then integrate these values with the original response values. The objective of this integration is to improve the feature discriminability of the neural feature response. Based on the calibration values, we propose a plugin-based calibration module incorporated into a modified ResNet architecture, termed Response Calibration Networks (ResCNet). Extensive experiments on datasets like CIFAR-10, CIFAR-100, SVHN, and ImageNet demonstrate the effectiveness of the proposed approach. The developed code is publicly available at https://github.com/tcmyxc/ResCNet.

Figures

Figures reproduced from arXiv: 2411.13582 by the authors.

Figure 1
Figure 1. Module architecture. From the module architecture, it can be seen that SENet further scales the features extracted by the ResNet residual branches, whereas our proposed architecture provides additional calibration of these features. In this section, we briefly present the feature response adjustment techniques related to this work, which are primarily categorized into regularization meth￾ods, activation functions, a… view at source ↗
Figure 2
Figure 2. The distribution of response values after model convergence using ResNet-32 on CIFAR-100 dataset. From the outer contour of the violin plot, it can be observed that the response values of a single convolutional kernel (neuron) roughly follow a Gaussian distribution. 3. Pre-analysis for Activation Responses We first obtained the features of the samples after passing through the feature extraction layer and then perfo… view at source ↗
Figure 3
Figure 3. ResNet block and ResCNet block. The output of the residual branch is then added to the shortcut branch and passed through a ReLU function. We integrate the proposed method (similar to the SE Layer) after the second BN layer. By examining the feature processing flow, we can observe the difference between our method and SE layer. SE layer scales the original features and uses them as the final output of the residual b… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Top-1 error of ImageNet using ResNet and ResCNet. From the training curves, it can be observed that the error rate of the ResCNet model is lower than that of the ResNet model. 5.2.2. Visualization of Calibration Effect We visualized the calibrated features and compared…
Figure 5
Figure 5. Figure 5: Comparison of response value distributions. We compared the response value distributions of the ResNet-32, SENet-32, and ResCNet-32 models on the CIFAR￾100 dataset. The red dashed bounding boxes serve as visual indicators of the enhanced discriminability in neural feat…
Figure 6
Figure 6. Figure 6: Proposed integration schemes. Sequential and parallel refer to the position of the calibration layer relative to the activation function layer. The original approach (ResCNet) is termed “serial scheme”. Experimental results show minimal differences in final model perfo…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 41 canonical work pages

  1. [1]

    Krizhevsky, I

    A. Krizhevsky, I. Sutskever, G. E. Hinton, Imagenet classification with deep convolutional neural networks, Advances in Neural Information Processing Systems 25 (2012)

    2012

  2. [2]

    Simonyan, A

    K. Simonyan, A. Zisserman, Very deep convolutional networks for large- scale image recognition, arXiv preprint arXiv:1409.1556 (2014)

    arXiv 2014

  3. [3]

    Szegedy, W

    C. Szegedy, W. Liu, Y. Jia, P. Sermanet, S. Reed, D. Anguelov, D. Er- han, V. Vanhoucke, A. Rabinovich, Going deeper with convolutions, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2015, pp. 1–9. 19

    work page 2015

  4. [4]

    K. He, X. Zhang, S. Ren, J. Sun, Deep residual learning for image recog- nition, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 770–778

    work page 2016

  5. [5]

    Huang, Z

    G. Huang, Z. Liu, L. Van Der Maaten, K. Q. Weinberger, Densely con- nected convolutional networks, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 4700–4708

    work page 2017

  6. [6]

    A. G. Howard, M. Zhu, B. Chen, D. Kalenichenko, W. Wang, T. Weyand, M. Andreetto, H. Adam, Mobilenets: Efficient convolu- tional neural networks for mobile vision applications, arXiv preprint arXiv:1704.04861 (2017)

    arXiv 2017

  7. [7]

    F. N. Iandola, S. Han, M. W. Moskewicz, K. Ashraf, W. J. Dally, K. Keutzer, Squeezenet: Alexnet-level accuracy with 50x fewer param- eters and ¡0.5 mb model size, arXiv preprint arXiv:1602.07360 (2016)

    arXiv 2016

  8. [8]

    Guariglia, Harmonic sierpinski gasket and applications, Entropy 20 (9) (2018) 714

    E. Guariglia, Harmonic sierpinski gasket and applications, Entropy 20 (9) (2018) 714

    work page 2018

Show all 54 references
  1. [9]

    Guariglia, Primality, fractality, and image analysis, Entropy 21 (3) (2019) 304

    E. Guariglia, Primality, fractality, and image analysis, Entropy 21 (3) (2019) 304

    2019

  2. [10]

    Guariglia, Fractional calculus, zeta functions and shannon entropy, Open Mathematics 19 (1) (2021) 87–100

    E. Guariglia, Fractional calculus, zeta functions and shannon entropy, Open Mathematics 19 (1) (2021) 87–100

    2021

  3. [11]

    L. Yang, H. Su, C. Zhong, Z. Meng, H. Luo, X. Li, Y. Y. Tang, Y. Lu, Hyperspectral image classification using wavelet transform-based smooth ordering, International Journal of Wavelets, Multiresolution and Information Processing 17 (06) (2019) 1950050

    2019

  4. [12]

    S. S. Basha, S. R. Dubey, V. Pulabaigari, S. Mukherjee, Impact of fully connected layers on performance of convolutional neural networks for image classification, Neurocomputing 378 (2020) 112–119

    2020

  5. [13]

    K. N. Qureshi, O. Kaiwartya, G. Jeon, F. Piccialli, Neurocomputing for internet of things: object recognition and detection strategy, Neurocom- puting 485 (2022) 263–273

    2022

  6. [14]

    L. Zhou, S. Zhang, T. Qiu, W. Xu, Z. Feng, M. Song, Patchdetector: Pluggable and non-intrusive patch for small object detection, Neuro- computing 589 (2024) 127715. 20

    2024

  7. [15]

    S. G. Mallat, A theory for multiresolution signal decomposition: the wavelet representation, IEEE Transactions on Pattern Analysis and Ma- chine Intelligence 11 (7) (1989) 674–693

    1989

  8. [16]

    Zheng, Y

    X. Zheng, Y. Y. Tang, J. Zhou, A framework of adaptive multiscale wavelet decomposition for signals on undirected graphs, IEEE Transac- tions on Signal Processing 67 (7) (2019) 1696–1711

    2019

  9. [17]

    R. C. Guido, Effectively interpreting discrete wavelet transformed sig- nals, IEEE Signal Processing Magazine 34 (3) (2017) 89–100

    2017

  10. [18]

    Guariglia, S

    E. Guariglia, S. Silvestrov, Fractional-wavelet analysis of positive def- inite distributions and wavelets on d’(c), in: Engineering mathematics II, Springer, 2016, pp. 337–353

    2016

  11. [19]

    Z. Meng, Q. Qian, M. Xu, B. Yu, A. R. Yıldız, S. Mirjalili, Pinn-form: A new physics-informed neural network for reliability analysis with par- tial differential equation, Computer Methods in Applied Mechanics and Engineering 414 (2023) 116172

    2023

  12. [20]

    M. Raissi, Forward–backward stochastic neural networks: deep learn- ing of high-dimensional partial differential equations, in: Peter Carr Gedenkschrift: Research Advances in Mathematical Finance, World Sci- entific, 2024, pp. 637–655

    2024

  13. [21]

    Muhammad, J

    K. Muhammad, J. Ahmad, S. W. Baik, Early fire detection using convo- lutional neural networks during surveillance for effective disaster man- agement, Neurocomputing 288 (2018) 30–42

    2018

  14. [22]

    Gupta, S

    A. Gupta, S. Watson, H. Yin, Deep learning-based aerial image segmen- tation with open data for disaster impact assessment, Neurocomputing 439 (2021) 22–33

    2021

  15. [23]

    H. Yu, L. T. Yang, Q. Zhang, D. Armstrong, M. J. Deen, Convolutional neural networks for medical image analysis: state-of-the-art, compar- isons, improvement and perspectives, Neurocomputing 444 (2021) 92– 110

    2021

  16. [24]

    Niyas, S

    S. Niyas, S. Pawan, M. A. Kumar, J. Rajan, Medical image segmentation with 3d convolutional neural networks: A survey, Neurocomputing 493 (2022) 397–413. 21

    2022

  17. [25]

    Ioffe, C

    S. Ioffe, C. Szegedy, Batch normalization: Accelerating deep network training by reducing internal covariate shift, in: International Confer- ence on Machine Learning, pmlr, 2015, pp. 448–456

    2015

  18. [26]

    J. Hu, L. Shen, G. Sun, Squeeze-and-excitation networks, in: Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recogni- tion, 2018, pp. 7132–7141

    2018

  19. [27]

    X. Chen, C. Liang, D. Huang, E. Real, K. Wang, H. Pham, X. Dong, T. Luong, C.-J. Hsieh, Y. Lu, et al., Symbolic discovery of optimiza- tion algorithms, Advances in neural information processing systems 36 (2024)

    2024

  20. [28]

    Dou, S.-C

    Z.-C. Dou, S.-C. Chu, Z. Zhuang, A. R. Yildiz, J.-S. Pan, Gbrun: A gradient search-based binary runge kutta optimizer for feature selection, Journal of Internet Technology 25 (3) (2024) 341–353

    2024

  21. [29]

    J. L. Ba, J. R. Kiros, G. E. Hinton, Layer normalization, arXiv preprint arXiv:1607.06450 (2016)

    2016 arXiv

  22. [30]

    Hendrycks, K

    D. Hendrycks, K. Gimpel, Gaussian error linear units (gelus), arXiv preprint arXiv:1606.08415 (2016)

    2016 arXiv

  23. [31]

    X. Li, W. Wang, X. Hu, J. Yang, Selective kernel networks, in: Proceed- ings of the IEEE Conference on Computer Vision and Pattern Recogni- tion, 2019, pp. 510–519

    2019

  24. [32]

    M. U. Erda¸ s, M. Kopar, B. S. Yildiz, A. R. Yildiz, Optimum design of a seat bracket using artificial neural networks and dandelion optimization algorithm, Materials Testing 65 (12) (2023) 1767–1775

    2023

  25. [33]

    S. M. Sait, P. Mehta, A. R. Yıldız, B. S. Yıldız, Optimal design of struc- tural engineering components using artificial neural network-assisted crayfish algorithm, Materials Testing (2024)

    2024

  26. [34]

    Y. Wu, K. He, Group normalization, in: Proceedings of the European Conference on Computer Vision (ECCV), 2018, pp. 3–19

    2018

  27. [35]

    Ulyanov, A

    D. Ulyanov, A. Vedaldi, V. Lempitsky, Instance normalization: The missing ingredient for fast stylization, arXiv preprint arXiv:1607.08022 (2016). 22

    2016 arXiv

  28. [36]

    S. Woo, J. Park, J.-Y. Lee, I. S. Kweon, Cbam: Convolutional block attention module, in: Proceedings of the European Conference on Com- puter Vision (ECCV), 2018, pp. 3–19

    2018

  29. [37]

    D. Li, J. Hu, C. Wang, X. Li, Q. She, L. Zhu, T. Zhang, Q. Chen, Invo- lution: Inverting the inherence of convolution for visual recognition, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2021, pp. 12321–12330

    2021

  30. [38]

    Vaswani, P

    A. Vaswani, P. Ramachandran, A. Srinivas, N. Parmar, B. Hechtman, J. Shlens, Scaling local self-attention for parameter efficient visual back- bones, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2021, pp. 12894–12904

    2021

  31. [39]

    Krizhevsky, G

    A. Krizhevsky, G. Hinton, Learning multiple layers of features from tiny images, Master’s thesis, Department of Computer Science, University of Toronto (2009)

    2009

  32. [40]

    Kingma, Auto-encoding variational bayes, arXiv preprint arXiv:1312.6114 (2013)

    D. Kingma, Auto-encoding variational bayes, arXiv preprint arXiv:1312.6114 (2013)

    2013 arXiv

  33. [41]

    Netzer, T

    Y. Netzer, T. Wang, A. Coates, A. Bissacco, B. Wu, A. Y. Ng, et al., Reading digits in natural images with unsupervised feature learning, in: NIPS workshop on deep learning and unsupervised feature learning, Granada, Spain, 2011, p. 7

    2011

  34. [42]

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, L. Fei-Fei, Imagenet: A large-scale hierarchical image database, in: 2009 IEEE Conference on Computer Vision and Pattern Recognition, Ieee, 2009, pp. 248–255

    2009

  35. [43]

    Loshchilov, F

    I. Loshchilov, F. Hutter, Sgdr: Stochastic gradient descent with warm restarts, arXiv preprint arXiv:1608.03983 (2016)

    2016 arXiv

  36. [44]

    T. He, Z. Zhang, H. Zhang, Z. Zhang, J. Xie, M. Li, Bag of tricks for image classification with convolutional neural networks, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2019, pp. 558–567

    2019

  37. [45]

    E. D. Cubuk, B. Zoph, D. Mane, V. Vasudevan, Q. V. Le, Autoaug- ment: Learning augmentation strategies from data, in: Proceedings of 23 the IEEE Conference on Computer Vision and Pattern Recognition, 2019, pp. 113–123

    2019

  38. [46]

    Zhong, L

    Z. Zhong, L. Zheng, G. Kang, S. Li, Y. Yang, Random erasing data augmentation, in: Proceedings of the AAAI conference on Artificial Intelligence, 2020, pp. 13001–13008

    2020

  39. [47]

    Zhang, M

    H. Zhang, M. Cisse, Y. N. Dauphin, D. Lopez-Paz, mixup: Beyond empirical risk minimization, arXiv preprint arXiv:1710.09412 (2017)

    2017 arXiv

  40. [48]

    S. Yun, D. Han, S. J. Oh, S. Chun, J. Choe, Y. Yoo, Cutmix: Regular- ization strategy to train strong classifiers with localizable features, in: Proceedings of the IEEE International Conference on Computer Vision, 2019, pp. 6023–6032

    2019

  41. [49]

    DeVries, G

    T. DeVries, G. W. Taylor, Improved regularization of convolutional neu- ral networks with cutout, arXiv preprint arXiv:1708.04552 (2017)

    2017 arXiv

  42. [50]

    Dosovitskiy, L

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, et al., An image is worth 16x16 words: Transformers for image recognition at scale, arXiv preprint arXiv:2010.11929 (2020)

    2020 arXiv

  43. [51]

    Z. Liu, Y. Lin, Y. Cao, H. Hu, Y. Wei, Z. Zhang, S. Lin, B. Guo, Swin transformer: Hierarchical vision transformer using shifted windows, in: Proceedings of the IEEE international conference on computer vision, 2021, pp. 10012–10022

    2021

  44. [52]

    Touvron, M

    H. Touvron, M. Cord, A. Sablayrolles, G. Synnaeve, H. J´ egou, Going deeper with image transformers, in: Proceedings of the IEEE interna- tional conference on computer vision, 2021, pp. 32–42

    2021

  45. [53]

    E. Page, Approximations to the cumulative normal function and its inverse for use on a pocket calculator, Journal of the Royal Statistical Society Series C: Applied Statistics 26 (1) (1977) 75–76

    1977

  46. [54]

    S. R. Bowling, M. T. Khasawneh, S. Kaewkuekool, B. R. Cho, A lo- gistic approximation to the cumulative normal distribution, Journal of Industrial Engineering and Management 2 (1) (2009) 114–127. 24

    2009

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.