Pith. sign in

REVIEW 4 major objections 6 minor 55 references

FSTA-SNN:Frequency-based Spatial-Temporal Attention Module for Spiking Neural Networks

T0 review · 4 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read DCT attention cuts SNN spike firing by a third, lifting accuracy.

desk verdict FSTA-SNN reports strong accuracy and spike-reduction gains for a plug-in SNN attention module, but a missing same-backbone ablation leaves the central claim less secure than the paper suggests. read the letter →

arxiv 2501.14744 v2 pith:SYL5T4FT submitted 2024-12-15 cs.NE cs.CVcs.LG

Kairong Yu , Tianqing Zhang , Hongwei Wang , Qi Xu This is my paper · ORCID
classification cs.NEcs.CVcs.LG
keywords spikingneuralnetworksfrequency-domainanalysisdiscretecosinetransformspatial-temporalattentionspikefiringratereductionenergy-efficientcomputingneuromorphicmodule
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 claims that spiking neural networks (SNNs) have consistent, architecture-independent learning preferences that can be read off the Fourier spectrum of their intermediate spike outputs: shallow layers emphasize vertical feature variations, deeper layers shift to horizontal variations, and the frequency content of a layer barely changes across time steps. On the strength of those observations, the authors design a plug-in Frequency-based Spatial-Temporal Attention (FSTA) module that uses a fixed discrete cosine transform (DCT) to extract full-spectrum spatial features and a small temporal attention submodule to reweight time steps. Adding FSTA to standard SNN backbones, they report higher top-1 accuracy on CIFAR-10/100, ImageNet, and CIFAR10-DVS than the baselines they compare against, while cutting total spike firing rate by about 33.99% at roughly constant energy cost. The paper's contribution is both a diagnostic claim about where SNN learning happens in frequency space and a practical mechanism that exploits that structure.

What carries the argument

The load-bearing identity is that global average pooling (GAP) equals the $(0,0)$ coefficient of the 2D discrete cosine transform up to the constant factor $H \times W$, which the paper proves by evaluating the DCT basis at $u=v=0$. From this it argues that conventional spatial attention built on GAP only sees the lowest frequency band, and it replaces that with a non-trainable convolution whose fixed weights are the DCT basis functions, extracting the full frequency spectrum of the temporally averaged feature map. This DCT-based spatial attention submodule is paired with a temporal attention submodule that pools along time with learned balances between average and max pooling, then weights each time step. The two submodules run in series with learnable scale factors, producing the FSTA output; the fixed DCT kernels keep the added parameter count and floating-point cost low.

What would settle it

Train the same ResNet and VGG SNNs on the same datasets with a handful of random seeds and training schedules, then compute the centered 2D DFT of intermediate spike outputs at each layer and time step; if the shallow-layer horizontal-axis concentration, the deep-layer vertical-axis shift, or the cross-time-step spectral overlap fails to appear consistently across seeds, architectures, or datasets, the claimed universal learning preference is contradicted.

Watch

Extended reading notes

Core claim

The central claim is that SNN intermediate spike outputs have a stable frequency signature: the Fourier spectrum of shallow-layer spikes is concentrated along the central horizontal axis, which the authors interpret as a preference for vertical feature variations, and as depth increases the spectrum shifts toward the vertical axis, indicating a preference for horizontal variations. Across time steps within a layer, the spectrum remains nearly identical except for amplitude, so the authors conclude that increasing simulation time adds little new feature information. They treat this spectral profile as a network-level learning preference that holds across ResNet and VGG architectures and across static and event-stream datasets, and use it to motivate a module that suppresses redundant frequency components while amplifying preferred ones.

Load-bearing premise

The load-bearing premise is that the Fourier spectra measured from the paper's trained networks are genuine, stable learning preferences of SNNs in general, rather than artifacts of the particular checkpoints, normalizations, or averaging choices used to draw Figure 1; if the shallow-to-deep vertical-to-horizontal shift and time-step invariance do not survive across seeds and architectures, the module's theoretical justification weakens even if its empirical gains persist.

Editorial extensions

If this is right

  • Plugging FSTA into ResNet34 lifts ImageNet top-1 accuracy to 70.23% at four time steps, up from 67.69% for the Real Spike baseline.
  • On CIFAR10-DVS, FSTA with ResNet20 reaches 82.70% at sixteen time steps, up from 78.70% for the MPBN baseline at ten steps.
  • The module reduces total spike firing rate by 33.99% across the network, so the accuracy gain comes with fewer spikes rather than more computation.
  • Because each layer's spectrum is stable across time steps, the spatial attention submodule can be shared across time, keeping the added parameter count minimal.
  • Global average pooling, the compression used by standard attention modules, is just the zero-frequency DCT coefficient; replacing it with fixed DCT kernels widens the frequency coverage of spatial attention.

Reading between the lines

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

  • The authors do not test whether the spectral learning preferences appear across random seeds and training schedules; if they do, layer-wise frequency regularization could push SNNs toward even lower firing rates without retraining the attention module.
  • The fixed DCT kernels imply the benefit comes largely from the attention mask shape rather than learned frequency filters; ablating the sigmoid and linear mapping would isolate how much of the gain is due to the frequency extraction itself.
  • The temporal-stability observation suggests that reducing the number of simulation steps, rather than just weighting them, may be a cheaper way to exploit the same insight; the paper does not experiment with step-count reduction.
  • If the shallow-vertical/deep-horizontal pattern generalizes to other spiking architectures or neuromorphic event datasets, frequency analysis could become a standard diagnostic for SNN layer design, though this remains an extension beyond the paper's evidence.
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

4 major / 6 minor

Summary. The paper proposes FSTA, a plug-and-play attention module for spiking neural networks, based on a frequency-domain analysis of intermediate spike outputs. The spatial submodule applies fixed DCT convolution kernels to extract full-spectrum features, while the temporal submodule uses average/max pooling with learnable parameters to rescale spike features across time steps. Experiments on CIFAR-10/100, ImageNet, and CIFAR10-DVS report accuracy improvements over published state-of-the-art results and a 33.99% reduction in spike firing rate, with a small claimed increase in computational cost. The paper also includes a proof that global average pooling corresponds to the lowest-frequency component of the 2D DCT.

Significance. If the reported gains are robust, FSTA is an inexpensive and architecture-agnostic addition that improves accuracy while reducing spike firing rate, and the GAP-as-lowest-frequency observation is a clean, checkable insight. The manuscript provides code, reports error bars on key results, and evaluates on both static and neuromorphic datasets. However, the central causal claim is currently supported only by comparisons against published numbers and by ablations that never remove the module entirely. The significance of the contribution is therefore conditional on controlled same-pipeline experiments being added.

major comments (4)
  1. [Tables 1–3 and Table 4] The paper's central claim is that inserting FSTA into standard SNN backbones improves accuracy and reduces firing rate, yet no experiment compares a given backbone trained with FSTA against the same backbone trained without FSTA under the identical pipeline. Tables 1–3 compare against published numbers from other papers, and the ablation in Table 4 only compares submodule combinations (modes a/b/c); it never removes FSTA entirely. Because training schedules, augmentations, normalization, and other recipe details can easily account for several accuracy points in the SNN literature, this omission is load-bearing. Please add a controlled ablation (e.g., ResNet20/19 on CIFAR-100, ResNet20 on CIFAR10-DVS, ResNet34 on ImageNet) with and without FSTA, reporting mean and standard deviation, and report the corresponding firing-rate comparison under the same recipe.
  2. [Equations (16)–(18)] The temporal attention submodule is not fully specified. With X in R^{T,C,H,W}, average/max pooling over spatial dimensions yield tensors in R^{T,C,1,1}, so M = alpha*favg + beta*fmax is in R^{T,C,1,1}, not R^{T,C} as written; Eq. (17) then averages over the temporal dimension to produce a C-dimensional vector, which cannot be linearly mapped to a T-dimensional weight vector Tw in Eq. (18). Please clarify the intended axes, tensor shapes, whether alpha/beta are per-channel or scalar, and how Tw is obtained. As written, the temporal mechanism cannot be reproduced from the equations alone.
  3. [Table 5 and Fig. 4] The energy and firing-rate claims are not quantitatively grounded. Table 5 reports ACs, MACs, FLOPs, and energy only for FSTA-equipped models, with no vanilla-SNN baseline, so the claim that the module does not significantly increase energy consumption is unverified. Figure 4 compares FSTA-SNN with a vanilla SNN, but it is not stated whether that vanilla SNN uses the same training recipe, initialization, and time steps; without this, the 33.99% firing-rate reduction cannot be attributed to the module. Please include the vanilla baseline in Table 5 and describe the energy-estimation methodology (e.g., per-AC and per-MAC energy constants) in the text.
  4. [Observations 1 and 2] The frequency-analysis observations that motivate the entire design are stated qualitatively. Terms such as 'remarkably similar', 'significant overlap', and 'gradually shifts' are not accompanied by quantitative measurements (e.g., spectral-energy ratios, correlation, or KL divergence between layers/time steps) or by a description of how the DFT magnitudes are normalized and averaged over the test set. Since the architectural choices in Eqs. (13)–(22) depend on these observations, please add quantitative support or explicitly reframe them as informal intuitions rather than empirical findings.
minor comments (6)
  1. [Equation (11)] The equality in Eq. (11) holds only up to the normalization constants of the DCT basis; please state that the result is proportional to the lowest-frequency component rather than exactly equal, or specify the DCT normalization convention used.
  2. [Observation 1] The text refers to Eq. (7) to justify the spectral behavior, but Eq. (7) is the IDFT; Eq. (8) or Eq. (10) seems intended.
  3. [Equation (22)] Scale_t and Scale_s are introduced but never defined; please state whether they are learnable, how they are initialized, and whether they are per-layer or global.
  4. [Table 4] Table 4 does not state the dataset, network, and time step in the caption; these details appear only in the text and should be moved into the table caption.
  5. [Section heading] The section heading 'Comparion with SOTA methods' contains a typo and should read 'Comparison with SOTA methods'.
  6. [Figure 1] The caption of Figure 1 should describe the preprocessing used to produce the spectra (e.g., which spike outputs are included, how magnitudes are normalized, how many samples are averaged, and the frequency-axis convention); as written, the analysis cannot be reproduced.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the FSTA derivation is self-contained; the GAP-as-lowest-frequency-DCT identity is an explicit proof, reported gains are benchmark-based, and self-citations are not load-bearing.

full rationale

The claimed derivation chain is not circular. The key first-principles step, that global average pooling is proportional to the lowest-frequency 2D DCT component, is proven directly in Eq. 11 by substituting u=v=0 into the DCT definition; this is an externally checkable mathematical identity, not an assumption that already contains the module's output. Observations 1 and 2 are qualitative empirical characterizations of spike spectra from trained networks, and the FSTA module is a design motivated by those observations rather than a quantity that is then 'predicted' back from them. The paper's accuracy and firing-rate results are measured on held-out benchmarks and compared against independent published SOTA methods, so no reported number is recovered from the motivating spectra by construction. Hyperparameters such as DCT kernel size are ablated against accuracy, and the submodule combination study compares different configurations; none of these choices fits a target result. The self-citations (e.g., Xu et al. 2023a,b) appear only in the related-work discussion and do not supply any load-bearing premise or uniqueness theorem for the FSTA design. The lack of a same-pipeline vanilla-SNN accuracy baseline is a genuine experimental-control weakness and should be noted as a correctness risk, but it is not an instance of circularity because no equation or fitted parameter in the paper reduces to its own input. Therefore no circular step can be exhibited under the standards required here.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical or ontological entities; FSTA is a neural network module built from existing transforms. The central claim rests on several free hyperparameters and on an interpretive assumption that spectral patterns reflect learning preferences.

free parameters (3)
  • DCT kernel size (Conv_dct frequency range) = 7x7
    Selected via a CIFAR-10 sweep (Table 6: 3x3=92.36%, 5x5=92.58%, 7x7=93.01%) and applied to all other datasets; controls how many DCT frequency components the spatial branch sees.
  • Scale_t and Scale_s in Eq. 22 = not reported
    Fusion scaling factors for temporal and spatial branches; they are learned or configured to balance enhancement and are not reported for any experiment.
  • Alpha and beta in Eq. 16 = not reported
    Learnable weights mixing average and max pooling in the temporal attention branch; initialization and final values are not given.
assumptions (4)
  • standard math DFT/DCT are lossless linear transforms with the standard convolution/basis properties used in Eqs. 6-10
    Invoked in Frequency Analysis and Methodology without proof; this is standard background.
  • ad hoc to paper The DFT magnitude of intermediate spike outputs is a valid proxy for what features SNNs learn
    Section 'Frequency Analysis' interprets spectral concentration as vertical/horizontal feature preferences; this interpretive link is asserted, not derived.
  • domain assumption LIF neuron dynamics (Eqs. 1-4) are the computational basis for all trained SNNs
    The method is evaluated on LIF-based networks; results may not transfer to other neuron models.
  • ad hoc to paper Suppressing low-energy spectral components and reducing spike firing rate removes redundancy without hurting task information
    This is the core premise of FSTA; it is supported only indirectly by final accuracy, not by an independent measure of task-relevant information.

how reviews work

0 comments
Cite this review

Pith. "Pith review of FSTA-SNN:Frequency-based Spatial-Temporal Attention Module for Spiking Neural Networks." pith.science (2026). https://pith.science/paper/SYL5T4FT

@misc{pith2026250114744,
  author       = {Pith},
  title        = {Pith review of: FSTA-SNN:Frequency-based Spatial-Temporal Attention Module for Spiking Neural Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SYL5T4FT}},
  note         = {Machine review of arXiv:2501.14744}
}
read the original abstract

Spiking Neural Networks (SNNs) are emerging as a promising alternative to Artificial Neural Networks (ANNs) due to their inherent energy efficiency. Owing to the inherent sparsity in spike generation within SNNs, the in-depth analysis and optimization of intermediate output spikes are often neglected. This oversight significantly restricts the inherent energy efficiency of SNNs and diminishes their advantages in spatiotemporal feature extraction, resulting in a lack of accuracy and unnecessary energy expenditure. In this work, we analyze the inherent spiking characteristics of SNNs from both temporal and spatial perspectives. In terms of spatial analysis, we find that shallow layers tend to focus on learning vertical variations, while deeper layers gradually learn horizontal variations of features. Regarding temporal analysis, we observe that there is not a significant difference in feature learning across different time steps. This suggests that increasing the time steps has limited effect on feature learning. Based on the insights derived from these analyses, we propose a Frequency-based Spatial-Temporal Attention (FSTA) module to enhance feature learning in SNNs. This module aims to improve the feature learning capabilities by suppressing redundant spike features.The experimental results indicate that the introduction of the FSTA module significantly reduces the spike firing rate of SNNs, demonstrating superior performance compared to state-of-the-art baselines across multiple datasets.

Figures

Figures reproduced from arXiv: 2501.14744 by the authors.

Figure 1
Figure 1. Comparison of the output spike frequency spec [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of the FSTA module and its internal submodules structure [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The different combinations of modules Mode ACs MACs Accuracy a 109.91M 38.93M 72.15% b 107.31M 38.93M 70.83% c 89.17M 41.74M 72.68% [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (2 more)
Figure 5
Figure 5. Figure 5: Visualization of the FSTA-SNN and vanilla SNN [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 4
Figure 4. Figure 4: Comparison of spike firing rates at various lay [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 30 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Ahmed, N.; Natarajan, T.; and Rao, K. R. 1974. Discrete cosine transform. IEEE transactions on Computers, 100(1): 90--93

    work page 1974

  4. [4]

    Brigham, E. O. 1988. The fast Fourier transform and its applications. Prentice-Hall, Inc

    work page 1988

  5. [5]

    S.; Gopinath, R

    Burrus, C. S.; Gopinath, R. A.; and Guo, H. 1998. Wavelets and wavelet transforms. rice university, houston edition, 98

    work page 1998

  6. [6]

    Chen, T.; Wang, L.; Li, J.; Duan, S.; and Huang, T. 2023. Improving spiking neural network with frequency adaptation for image classification. IEEE Transactions on Cognitive and Developmental Systems

    work page 2023

  7. [7]

    Deng, J.; Dong, W.; Socher, R.; Li, L.-J.; Li, K.; and Fei-Fei, L. 2009. ImageNet : A large-scale hierarchical image database. In 2009 IEEE Conference on Computer Vision and Pattern Recognition , 248--255. ISSN: 1063-6919

    work page 2009

  8. [8]

    Deng, L.; Wu, Y.; Hu, X.; Liang, L.; Ding, Y.; Li, G.; Zhao, G.; Li, P.; and Xie, Y. 2020. Rethinking the performance comparison between SNNS and ANNS. Neural networks, 121: 294--307

    work page 2020

Show all 55 references
  1. [9]

    Deng, L.; Wu, Y.; Hu, Y.; Liang, L.; Li, G.; Hu, X.; Ding, Y.; Li, P.; and Xie, Y. 2021. Comprehensive SNN Compression Using ADMM Optimization and Activity Regularization. Institute of Electrical and Electronics Engineers (IEEE), (99)

    2021

  2. [10]

    Deng, S.; Li, Y.; Zhang, S.; and Gu, S. 2022. Temporal efficient training of spiking neural network via gradient re-weighting. arXiv preprint arXiv:2202.11946

    2022 arXiv

  3. [11]

    Duan, C.; Ding, J.; Chen, S.; Yu, Z.; and Huang, T. 2022. Temporal effective batch normalization in spiking neural networks. Advances in Neural Information Processing Systems, 35: 34377--34390

    2022

  4. [12]

    K.; Ward, M.; Neftci, E

    Eshraghian, J. K.; Ward, M.; Neftci, E. O.; Wang, X.; Lenz, G.; Dwivedi, G.; Bennamoun, M.; Jeong, D. S.; and Lu, W. D. 2023. Training spiking neural networks using lessons from deep learning. Proceedings of the IEEE

    2023

  5. [13]

    M.; Devienne, P.; and Boulet, P

    Falez, P.; Tirilly, P.; Bilasco, I. M.; Devienne, P.; and Boulet, P. 2018. Mastering the output frequency in spiking neural networks. In 2018 international joint conference on neural networks (IJCNN), 1--8. IEEE

    2018

  6. [14]

    Fang, W.; Yu, Z.; Chen, Y.; Huang, T.; Masquelier, T.; and Tian, Y. 2021. Deep residual learning in spiking neural networks. Advances in Neural Information Processing Systems, 34: 21056--21069

    2021

  7. [15]

    S.; and Roy, K

    Garg, I.; Chowdhury, S. S.; and Roy, K. 2021. Dct-snn: Using dct to distribute spatial information over time for low-latency spiking neural networks. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 4671--4680

    2021

  8. [16]

    R.; Cheng, M.-M.; and Hu, S.-M

    Guo, M.-H.; Xu, T.-X.; Liu, J.-J.; Liu, Z.-N.; Jiang, P.-T.; Mu, T.-J.; Zhang, S.-H.; Martin, R. R.; Cheng, M.-M.; and Hu, S.-M. 2022 a . Attention mechanisms in computer vision: A survey. Computational visual media, 8(3): 331--368

    2022

  9. [17]

    Guo, S.; Yong, H.; Zhang, X.; Ma, J.; and Zhang, L. 2023 a . Spatial-frequency attention for image denoising. arXiv preprint arXiv:2302.13598

    2023 arXiv

  10. [18]

    Guo, Y.; Chen, Y.; Zhang, L.; Liu, X.; Wang, Y.; Huang, X.; and Ma, Z. 2022 b . IM-loss: information maximization loss for spiking neural networks. Advances in Neural Information Processing Systems, 35: 156--166

    2022

  11. [19]

    Guo, Y.; Peng, W.; Chen, Y.; Zhang, L.; Liu, X.; Huang, X.; and Ma, Z. 2023 b . Joint a-snn: Joint training of artificial and spiking neural networks via self-distillation and weight factorization. Pattern Recognition, 142: 109639

    2023

  12. [20]

    Guo, Y.; Tong, X.; Chen, Y.; Zhang, L.; Liu, X.; Ma, Z.; and Huang, X. 2022 c . Recdis-snn: Rectifying membrane potential distribution for directly training spiking neural networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 326--335

    2022

  13. [21]

    Guo, Y.; Zhang, L.; Chen, Y.; Tong, X.; Liu, X.; Wang, Y.; Huang, X.; and Ma, Z. 2022 d . Real spike: Learning real-valued spikes for spiking neural networks. In European Conference on Computer Vision, 52--68. Springer

    2022

  14. [22]

    Guo, Y.; Zhang, Y.; Chen, Y.; Peng, W.; Liu, X.; Zhang, L.; Huang, X.; and Ma, Z. 2023 c . Membrane potential batch normalization for spiking neural networks. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 19420--19430

    2023

  15. [23]

    Han, B.; Srinivasan, G.; and Roy, K. 2020. Rmp-snn: Residual membrane potential neuron for enabling deeper high-accuracy and low-latency spiking neural network. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 13558--13567

    2020

  16. [24]

    E.; Mohamed, A.-r.; Jaitly, N.; Senior, A.; Vanhoucke, V.; Nguyen, P.; Sainath, T

    Hinton, G.; Deng, L.; Yu, D.; Dahl, G. E.; Mohamed, A.-r.; Jaitly, N.; Senior, A.; Vanhoucke, V.; Nguyen, P.; Sainath, T. N.; et al. 2012. Deep neural networks for acoustic modeling in speech recognition: The shared views of four research groups. IEEE Signal processing magazin...

    2012

  17. [25]

    Hu, J.; Shen, L.; and Sun, G. 2018. Squeeze-and-excitation networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, 7132--7141

    2018

  18. [26]

    Huang, Z.; Zhang, Z.; Lan, C.; Zha, Z.-J.; Lu, Y.; and Guo, B. 2023. Adaptive frequency filters as efficient global token mixers. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 6049--6059

    2023

  19. [27]

    Kong, L.; Dong, J.; Ge, J.; Li, M.; and Pan, J. 2023. Efficient frequency domain-based transformers for high-quality image deblurring. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 5886--5895

    2023

  20. [28]

    Krizhevsky, A.; Nair, V.; and Hinton, G. 2010. Cifar-10 (canadian institute for advanced research). URL http://www. cs. toronto. edu/kriz/cifar. html, 5(4): 1

    2010

  21. [29]

    Krizhevsky, A.; Sutskever, I.; and Hinton, G. E. 2012. Imagenet classification with deep convolutional neural networks. Advances in neural information processing systems, 25

    2012

  22. [30]

    Li, G.; Fang, Q.; Zha, L.; Gao, X.; and Zheng, N. 2022. HAM: Hybrid attention module in deep convolutional neural networks for image classification. Pattern Recognition, 129: 108785

    2022

  23. [31]

    Li, H.; Liu, H.; Ji, X.; Li, G.; and Shi, L. 2017. CIFAR10 - DVS : An Event - Stream Dataset for Object Classification . Frontiers in Neuroscience, 11

    2017

  24. [32]

    Li, Y.; Guo, Y.; Zhang, S.; Deng, S.; Hai, Y.; and Gu, S. 2021. Differentiable spike: Rethinking gradient-descent for training spiking neural networks. Advances in Neural Information Processing Systems, 34: 23426--23439

    2021

  25. [33]

    Ma, D.; Shen, J.; Gu, Z.; Zhang, M.; Zhu, X.; Xu, X.; Xu, Q.; Shen, Y.; and Pan, G. 2017. Darwin: A neuromorphic hardware co-processor based on spiking neural networks. Journal of systems architecture, 77: 43--51

    2017

  26. [34]

    Meng, Q.; Xiao, M.; Yan, S.; Wang, Y.; Lin, Z.; and Luo, Z.-Q. 2022. Training high-performance low-latency spiking neural networks by differentiation on spike representation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 12444--12453

    2022

  27. [35]

    N.; Namboodiri, V

    Patro, B. N.; Namboodiri, V. P.; and Agneeswaran, V. S. 2023. SpectFormer: Frequency and Attention is what you need in a Vision Transformer. arXiv preprint arXiv:2304.06446

    2023 arXiv

  28. [36]

    Pei, J.; Deng, L.; Song, S.; Zhao, M.; Zhang, Y.; Wu, S.; Wang, G.; Zou, Z.; Wu, Z.; He, W.; et al. 2019. Towards artificial general intelligence with hybrid Tianjic chip architecture. Nature, 572(7767): 106--111

    2019

  29. [37]

    Qin, Z.; Zhang, P.; Wu, F.; and Li, X. 2021. Fcanet: Frequency channel attention networks. In Proceedings of the IEEE/CVF international conference on computer vision, 783--792

    2021

  30. [38]

    Rathi, N.; Srinivasan, G.; Panda, P.; and Roy, K. 2020. Enabling deep spiking neural networks with hybrid conversion and spike timing dependent backpropagation. arXiv preprint arXiv:2005.01807

    2020 arXiv

  31. [39]

    Ronneberger, O.; Fischer, P.; and Brox, T. 2015. U-net: Convolutional networks for biomedical image segmentation. In Medical image computing and computer-assisted intervention--MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceedings, part II...

    2015

  32. [40]

    Roy, K.; Jaiswal, A.; and Panda, P. 2019. Towards spike-based machine intelligence with neuromorphic computing. Nature, 575(7784): 607--617

    2019

  33. [41]

    D.; Kulkarni, S

    Schuman, C. D.; Kulkarni, S. R.; Parsa, M.; Mitchell, J. P.; Kay, B.; et al. 2022. Opportunities for neuromorphic computing algorithms and applications. Nature Computational Science, 2(1): 10--19

    2022

  34. [42]

    Sengupta, A.; Ye, Y.; Wang, R.; Liu, C.; and Roy, K. 2019. Going deeper in spiking neural networks: VGG and residual architectures. Frontiers in neuroscience, 13: 95

    2019

  35. [43]

    K.; Wang, Y.; Pan, G.; and Tang, H

    Shen, J.; Xu, Q.; Liu, J. K.; Wang, Y.; Pan, G.; and Tang, H. 2023. Esl-snns: An evolutionary structure learning strategy for spiking neural networks. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, 86--93

    2023

  36. [44]

    Wu, J.; Xu, C.; Han, X.; Zhou, D.; Zhang, M.; Li, H.; and Tan, K. C. 2021. Progressive tandem learning for pattern recognition with deep spiking neural networks. IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(11): 7824--7840

    2021

  37. [45]

    Xu, Q.; Gao, Y.; Shen, J.; Li, Y.; Ran, X.; Tang, H.; and Pan, G. 2024. Enhancing adaptive history reserving by spiking convolutional block attention module in recurrent neural networks. Advances in Neural Information Processing Systems, 36

    2024

  38. [46]

    K.; Tang, H.; and Pan, G

    Xu, Q.; Li, Y.; Fang, X.; Shen, J.; Liu, J. K.; Tang, H.; and Pan, G. 2023 a . Biologically inspired structure learning with reverse knowledge distillation for spiking neural networks. arXiv preprint arXiv:2304.09500

    2023 arXiv

  39. [47]

    K.; Tang, H.; and Pan, G

    Xu, Q.; Li, Y.; Shen, J.; Liu, J. K.; Tang, H.; and Pan, G. 2023 b . Constructing deep spiking neural networks from artificial neural networks with knowledge distillation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 7886--7895

    2023

  40. [48]

    Yang, Q.; Wu, J.; Zhang, M.; Chua, Y.; Wang, X.; and Li, H. 2022. Training spiking neural networks with local tandem learning. Advances in Neural Information Processing Systems, 35: 12662--12676

    2022

  41. [49]

    Yao, M.; Gao, H.; Zhao, G.; Wang, D.; Lin, Y.; Yang, Z.; and Li, G. 2021. Temporal-wise attention spiking neural networks for event streams classification. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 10221--10230

    2021

  42. [50]

    Yao, M.; Hu, J.; Zhao, G.; Wang, Y.; Zhang, Z.; Xu, B.; and Li, G. 2023 a . Inherent redundancy in spiking neural networks. In Proceedings of the IEEE/CVF international conference on computer vision, 16924--16934

    2023

  43. [51]

    Yao, M.; Zhao, G.; Zhang, H.; Hu, Y.; Deng, L.; Tian, Y.; Xu, B.; and Li, G. 2023 b . Attention spiking neural networks. IEEE transactions on pattern analysis and machine intelligence, 45(8): 9393--9410

    2023

  44. [52]

    Yao, X.; Li, F.; Mo, Z.; and Cheng, J. 2022. Glif: A unified gated leaky integrate-and-fire neuron for spiking neural networks. Advances in Neural Information Processing Systems, 35: 32160--32171

    2022

  45. [53]

    Yin, B.; Corradi, F.; and Boht \'e , S. M. 2021. Accurate and efficient time-domain classification with adaptive spiking recurrent neural networks. Nature Machine Intelligence, 3(10): 905--913

    2021

  46. [54]

    Zheng, H.; Wu, Y.; Deng, L.; Hu, Y.; and Li, G. 2021. Going deeper with directly-trained larger spiking neural networks. In Proceedings of the AAAI conference on artificial intelligence, volume 35, 11062--11070

    2021

  47. [55]

    Zhu, R.-J.; Zhang, M.; Zhao, Q.; Deng, H.; Duan, Y.; and Deng, L.-J. 2024. Tcja-snn: Temporal-channel joint attention for spiking neural networks. IEEE Transactions on Neural Networks and Learning Systems

    2024

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.