Pith. sign in

REVIEW 3 major objections 5 minor 41 references

MSVIT: Improving Spiking Vision Transformer Using Multi-scale Attention Fusion

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A spiking vision transformer with multi-scale attention fusion reaches 85.06% top-1 accuracy on ImageNet-1K, surpassing the previous SNN-transformer state of the art.

desk verdict A plausible, incremental SNN transformer paper whose reported gains likely hold but whose central attribution claim is not proven by the current evidence. read the letter →

arxiv 2505.14719 v3 pith:CXSZUQHC submitted 2025-05-19 cs.CV cs.AI

Wei Hua , Chenlin Zhou , Jibin Wu , Yansong Chua , Yangyang Shu This is my paper · ORCID
classification cs.CVcs.AI
keywords spikingneuralnetworksvisiontransformermulti-scaleattentionspike-drivenimageclassificationneuromorphicdatasetsenergyefficiencyhierarchicalarchitecture
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper argues that the performance gap between spiking and conventional vision transformers comes in part from a single-scale bottleneck in spiking attention, and that a spike-compatible multi-scale fusion can close part of that gap. It introduces MSVIT, a hierarchical spiking transformer whose attention heads summarize feature projections at different scales by column sums, add those summaries, and gate the value spikes with the result. Directly trained from scratch on ImageNet-1K with 69.80M parameters and four time steps, MSVIT reports 85.06% top-1 accuracy, surpassing QKFormer's 84.22% and, the authors claim, becoming the state of the art among SNN-transformer architectures. The same recipe improves CIFAR and neuromorphic benchmarks, so the paper's claim is that multi-scale attention fusion helps across static and event-driven vision tasks.

What carries the argument

The load-bearing mechanism is MSSA, defined as $\text{MSSA}(Q,P,V) = \text{SN}(\text{SUM}_c(Q) \oplus \text{SUM}_c(P)) \otimes V$, where $\text{SUM}_c$ is a column-wise summation that turns each spike-form projection into an $N \times 1$ attention vector. The fusion of the low-level projection Q (from a $1\times1$ convolution) with the high-level projection P (from a $3\times3$ convolution) replaces the $QK^T$ matrix multiplication, reducing attention complexity from $O(N^2 D)$ to $O(ND)$. The surrounding design—SPEMSF's multi-scale patch embedding, the hybrid placement of MSSA in early stages and standard spiking self-attention in the final stage, and the hierarchical token downsampling—supports but does not itself carry the novelty claim.

What would settle it

Train the same MSVIT architecture on ImageNet-1K but replace the MSSA column-sum fusion in stages 1-2 with the standard spiking self-attention, or with a concatenation-based fusion, while keeping SPEMSF, stage depths, and the training recipe fixed; if top-1 accuracy stays around 85%, the multi-scale fusion is not the cause of the reported gain.

Watch

Extended reading notes

Core claim

According to the paper, the central discovery is that multi-scale spiking attention (MSSA) can replace the dot-product attention used in spiking transformers with a column-sum fusion that preserves spike-driven computation and linear complexity while improving accuracy. Each attention head computes two spike-form projections, Q and P, from convolutional paths with different receptive fields; summing over the feature dimension gives per-token attention scores in spike form, their element-wise sum gates V. The paper claims this fusion of low-level and high-level features enriches spike representations and, together with a hierarchical three-stage design and a multi-scale spiking patch embedding (SPEMSF), lifts ImageNet-1K top-1 accuracy to 85.06%, exceeding QKFormer and earlier SNN transformers.

Load-bearing premise

The central claim rests on the assumption that fusing the attention vectors from two different image scales is the reason accuracy improves, rather than the other changes introduced at the same time—the new patch embedding, the hierarchical stages, and where the two attention types are placed.

Editorial extensions

If this is right

  • If the ImageNet-1K result is reproducible, directly trained SNN transformers can reach roughly 85% top-1 accuracy, within striking distance of conventional ANN transformers while using spike-based accumulate operations and a fraction of the estimated energy.
  • QKFormer-style Q-K attention is not necessary for a strong spiking transformer; fusing two spike projections by column sums can match or exceed it.
  • Linear-complexity attention of this form would let spiking transformers process higher-resolution inputs or longer token sequences without the quadratic cost of standard self-attention.
  • On CIFAR100 the model surpasses a ViT baseline, suggesting the multi-scale fusion helps close part of the SNN-ANN gap on smaller static datasets as well as on large-scale ImageNet-1K.

Reading between the lines

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

  • The column-sum fusion effectively removes token-token interactions from attention; if the reported gain is real, it would imply that diverse input projections matter more than pairwise similarity for spiking transformers—an inference the paper does not state.
  • Because the ablation varies only which projections (Q, P, or both) are fused and not the new patch embedding, hierarchical stages, or hybrid attention placement, the accuracy gain may be partly caused by those components; a controlled comparison with SPEMSF added to QKFormer would isolate MSSA's contribution.
  • The fusion mechanism is a natural candidate for temporal multi-scale attention in event-stream tasks, where summing over time windows could play the same role as summing over scale, though the paper tests only spatial scales.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. MSVIT is a hierarchical spiking vision transformer for image and neuromorphic classification. It introduces Multi-Scale Spiking Attention (MSSA), which computes per-token scalar values by column-summing two spike-form projections Q and P, adding them, passing through a spiking neuron, and element-wise multiplying with V. It also introduces a Spiking Patch Embedding with Multi-Scale Feature Fusion (SPEMSF) and a three-stage hierarchy that uses MSSA in stages 1-2 and standard spiking self-attention (SSA) in stage 3. The authors report 85.06% top-1 accuracy on ImageNet-1K (69.80M parameters, 224x224 input, 4 time steps, trained from scratch), surpassing QKFormer by 0.84-1.29% across three model sizes, and they claim state-of-the-art status among SNN transformer architectures. Additional results on CIFAR10/100, DVS128-Gesture, and CIFAR10-DVS are provided, along with an ablation study on CIFAR100 that varies which projections feed the column-sum operation.

Significance. If the reported ImageNet result is reproducible, it is a strong empirical data point for spiking vision transformers, narrowing the gap to ANN-based transformers while maintaining low theoretical energy consumption. The code is public and the energy estimates follow standard 45nm assumptions. However, the paper's central novelty claim—that the accuracy gain comes from multi-scale attention fusion in Eq. (7)—is not supported by the experiments as reported. The mechanism is a per-token gating operation rather than a cross-token attention mechanism, and the ablation does not isolate it from SPEMSF, the hierarchical design, or the hybrid attention placement. The significance of the work therefore depends on additional controlled experiments that are currently missing.

major comments (3)
  1. [§3.2, Eq. (7)-(8)] The operation in Eq. (7) is per-token gating, not multi-scale attention. With pointwise or small-kernel convolutions producing Q, P, and V, SUM_c collapses the feature dimension to one scalar per token; the output is MSSA[n,:] = SN(alpha_q[n] + alpha_p[n]) * V[n,:], which scales V[n,:] without any Q-K style interaction across tokens. The paper describes this as 'multi-scale interactions' and 'attention scores,' which overstates the mechanism. Unless the projections in Eqs. (4)-(5) are explicitly spatial convolutions with multi-scale kernels—and Section 3.2 does not specify this—the module does not implement attention over tokens. Please clarify what cross-token or multi-scale information is actually fused, or rename the module and adjust the novelty claims accordingly.
  2. [Table 3 and §5] The ablation does not isolate MSSA. It only changes which sources (P, Q, P+P, Q+Q, P+Q) feed the column-sum, while keeping SPEMSF, the three-stage hierarchy, and the MSSA/SSA placement fixed. No baseline removes MSSA entirely or replaces it with SSA in stages 1-2, and no baseline removes SPEMSF. As a result, the ImageNet-1K gains over QKFormer in Table 1 (0.84-1.29%) cannot be attributed to Eq. (7); they could come from the convolutional SPEMSF, the hierarchical architecture, or the larger parameter counts. The Appendix E firing-rate comparison (0.306 vs 0.285) is descriptive and does not resolve this attribution. Additionally, the prose misreports Table 3 values: the text cites 81.25 and 81.15 for rows that read 81.36 and 81.56, respectively.
  3. [Table 1] The state-of-the-art claim is based on a limited baseline set and single-run numbers. The comparison omits SpikingResformer (cited in the Related Work) and other recent SNN transformers, and no variance or seed information is reported. Given that the margin over QKFormer is under 1.3 percentage points, the reported differences may not be statistically meaningful. Please add multiple runs with confidence intervals and expand the baseline table, or soften the SOTA claim.
minor comments (5)
  1. [§4.2] The citation for Spikformer is given as [Yao et al., 2022]; it should be [Zhou et al., 2023], as Spikformer is introduced by Zhou et al. and the reference list correctly attributes it to them.
  2. [Table 2] There are typos in the table and surrounding text: 'Comparision' should be 'Comparison' and 'Transfomer' should be 'Transformer'; the table formatting should be cleaned up to avoid overlapping values.
  3. [§3.1 and Appendix E] The channel setting 'C is set to 2' is ambiguous; please specify whether this is a channel multiplier or an absolute width, since the same phrasing appears in both the main text and the limitation appendix.
  4. [§3.2 and Table 3] The projections Q and P are described as 'convolutional layers' in Section 3.2, but the exact kernel sizes are only given in Table 3 as 1x1 and 3x1 convolutions. Please state the kernel sizes and whether these convolutions operate across spatial positions in the main text.
  5. [Appendix E] The statement that a higher firing rate (0.306 vs 0.285) implies that neurons 'transmit more information' needs support; firing rate alone is not a direct measure of information content, and the comparison does not control for architecture differences.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: MSVIT's claims are empirical and benchmarked against external results; no prediction reduces to a fitted input or self-citation chain.

full rationale

MSVIT is an empirical architecture paper. The central claims—85.06% ImageNet-1K top-1 accuracy and state-of-the-art status among SNN-transformers—are direct measurements from training runs compared with published baseline numbers, not quantities derived from fitted parameters or from the paper's own definitions. MSSA is defined by Eqs. (4)-(8), but the reported accuracy is not obtained by plugging those equations into themselves; the equations are an architectural proposal whose effect is tested on held-out test sets. The one self-citation (adopting the QKFormer training recipe, Sec. 4.2) is methodological and non-load-bearing: it does not force the ImageNet result. The ablation in Table 3 is incomplete as attribution evidence, because it varies only the Q/P fusion inputs and never removes MSSA or SPEMSF, and the mechanism may not perform true cross-token attention; however, these are confounding and novelty concerns rather than circularity. Appendix E even concedes that further analysis is needed to clarify how multi-scale processing influences spiking responses, which supports the view that the gain attribution is under-analyzed rather than definitionally forced. No equation in the paper reduces to its own input, no fitted constant is relabeled as a prediction, and no load-bearing argument rests on an unverified self-citation.

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

The central claim rests on standard SNN training assumptions (LIF dynamics, surrogate gradient), the energy estimation model, and the design choice that fusing column-wise sums of Q and P creates useful multi-scale attention. These are not derived in the paper.

free parameters (5)
  • Stage layer configuration = {1,2,7} (ImageNet), {1,1,2} (CIFAR), {0,1,1} (neuromorphic)
    Chosen by hand with ablation; directly determines model capacity and the reported accuracy.
  • Stage channel widths = C = 2, 2, 4
    Set per stage; authors note this increases parameters; selected to balance accuracy and size.
  • Q and P projection kernel sizes = 1x1 and 3x1 (Table 3); 3x3 in SPEMSF
    The multi-scale distinction is implemented via kernel size, but the main text equations leave the kernels unspecified.
  • Hybrid attention placement = MSSA in stages 1-2, SSA in stage 3
    Selected from the ablation in Table 3; hybrid beats all-MSSA and all-SSA on CIFAR100 and is transferred to ImageNet.
  • Timesteps T = 4 (static), 10/16 (neuromorphic)
    Standard SNN simulation length; accuracy and energy scale with T.
assumptions (4)
  • domain assumption LIF neuron model with surrogate gradient permits end-to-end training of SNNs
    All training uses this approximation; the paper does not justify it beyond citing prior SNN practice.
  • domain assumption The 45nm CMOS energy model (E_MAC=4.6pJ, E_AC=0.9pJ) is a valid basis for comparing model efficiency
    Appendix C: all energy numbers and efficiency comparisons rest on this hardware model; actual chip energy can differ.
  • ad hoc to paper Column-wise summation of Q and P, followed by element-wise addition and application to V, produces an attention vector that captures multi-scale feature relationships
    Core of MSSA; no theoretical derivation, only the empirical ablation in Table 3.
  • domain assumption The reported accuracy differences are meaningful despite single training runs
    No seeds or error bars are reported; the paper assumes the single-run numbers are representative.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MSVIT: Improving Spiking Vision Transformer Using Multi-scale Attention Fusion." pith.science (2026). https://pith.science/paper/CXSZUQHC

@misc{pith2026250514719,
  author       = {Pith},
  title        = {Pith review of: MSVIT: Improving Spiking Vision Transformer Using Multi-scale Attention Fusion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CXSZUQHC}},
  note         = {Machine review of arXiv:2505.14719}
}
read the original abstract

The combination of Spiking Neural Networks (SNNs) with Vision Transformer architectures has garnered significant attention due to their potential for energy-efficient and high-performance computing paradigms. However, a substantial performance gap still exists between SNN-based and ANN-based transformer architectures. While existing methods propose spiking self-attention mechanisms that are successfully combined with SNNs, the overall architectures proposed by these methods suffer from a bottleneck in effectively extracting features from different image scales. In this paper, we address this issue and propose MSVIT. This novel spike-driven Transformer architecture firstly uses multi-scale spiking attention (MSSA) to enhance the capabilities of spiking attention blocks. We validate our approach across various main datasets. The experimental results show that MSVIT outperforms existing SNN-based models, positioning itself as a state-of-the-art solution among SNN-transformer architectures. The codes are available at https://github.com/Nanhu-AI-Lab/MSViT.

Figures

Figures reproduced from arXiv: 2505.14719 by the authors.

Figure 1
Figure 1. Overview of MSVIT, a hierarchical spiking transformer with multi-scale spiking attention. Note C denotes the spike-form dimension. SDT-v2 [Yao et al., 2024a], which extended the Spike-driven Transformer into a meta form. On ImageNet-1K, SDT-v2 achieved top-1 accuracy up to 80.0% with 55 (M) parameters, surpassing SDT-v1 by 3.7%. Zhou et al. proposed QKFormer [Zhou et al., 2024a], which enhanced the spiking transform… view at source ↗
Figure 2
Figure 2. Comparison of three self-attention computation paradigms. (a) VSA employs floating-point matrix multiplication to assess the [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

41 extracted references · 27 canonical work pages

  1. [1]

    A low power, fully event- based gesture recognition system

    [Amiret al., 2017 ] Arnon Amir, Brian Taba, David Berg, Timothy Melano, Jeffrey McKinstry, Carmelo Di Nolfo, Tapan Nayak, Alexander Andreopoulos, Guillaume Gar- reau, Marcela Mendoza, et al. A low power, fully event- based gesture recognition system. InProceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 7243–7252,

    work page 2017

  2. [3]

    Is space-time attention all you need for video understanding? InICML, volume 2, page 4,

    [Bertasiuset al., 2021 ] Gedas Bertasius, Heng Wang, and Lorenzo Torresani. Is space-time attention all you need for video understanding? InICML, volume 2, page 4,

    2021

  3. [8]

    An image is worth 16x16 words: Transformers for image recognition at scale

    [Dosovitskiy, 2020] Alexey Dosovitskiy. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929,

    arXiv 2020

  4. [9]

    Multiscale vision transformers

    [Fanet al., 2021 ] Haoqi Fan, Bo Xiong, Karttikeya Man- galam, Yanghao Li, Zhicheng Yan, Jitendra Malik, and Christoph Feichtenhofer. Multiscale vision transformers. InProceedings of the IEEE/CVF international conference on computer vision, pages 6824–6835,

    work page 2021

  5. [10]

    Levit: a vision trans- former in convnet’s clothing for faster inference

    [Grahamet al., 2021 ] Benjamin Graham, Alaaeldin El- Nouby, Hugo Touvron, Pierre Stock, Armand Joulin, Herv´e J ´egou, and Matthijs Douze. Levit: a vision trans- former in convnet’s clothing for faster inference. InPro- ceedings of the IEEE/CVF international conference on computer vision, pages 12259–12269,

    work page 2021

  6. [11]

    Multi-scale self- attention for text classification

    [Guoet al., 2020 ] Qipeng Guo, Xipeng Qiu, Pengfei Liu, Xiangyang Xue, and Zheng Zhang. Multi-scale self- attention for text classification. InProceedings of the AAAI conference on artificial intelligence, volume 34, pages 7847–7854,

    work page 2020

  7. [12]

    Pct: Point cloud transformer.Computational Visual Me- dia, 7:187–199,

    [Guoet al., 2021 ] Meng-Hao Guo, Jun-Xiong Cai, Zheng- Ning Liu, Tai-Jiang Mu, Ralph R Martin, and Shi-Min Hu. Pct: Point cloud transformer.Computational Visual Me- dia, 7:187–199,

    work page 2021

  8. [13]

    Masked au- toencoders are scalable vision learners

    [Heet al., 2022 ] Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Doll´ar, and Ross Girshick. Masked au- toencoders are scalable vision learners. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 16000–16009,

    2022

Show all 41 references
  1. [19]

    Cifar10-dvs: an event-stream dataset for object classification.Frontiers in neuroscience, 11:309,

    [Liet al., 2017 ] Hongmin Li, Hanchao Liu, Xiangyang Ji, Guoqi Li, and Luping Shi. Cifar10-dvs: an event-stream dataset for object classification.Frontiers in neuroscience, 11:309,

    2017

  2. [20]

    Rethinking vision transformers for mo- bilenet size and speed

    [Liet al., 2023 ] Yanyu Li, Ju Hu, Yang Wen, Georgios Evan- gelidis, Kamyar Salahi, Yanzhi Wang, Sergey Tulyakov, and Jian Ren. Rethinking vision transformers for mo- bilenet size and speed. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 16889–16900,

    2023

  3. [21]

    Swin transformer: Hierarchical vision transformer using shifted windows

    [Liuet al., 2021 ] Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: Hierarchical vision transformer using shifted windows. InProceedings of the IEEE/CVF in- ternational conference on computer vision, pages 10012– 10022,

    2021

  4. [22]

    Ecoformer: Energy-saving atten- tion with linear complexity.Advances in Neural Informa- tion Processing Systems, 35:10295–10308,

    [Liuet al., 2022 ] Jing Liu, Zizheng Pan, Haoyu He, Jianfei Cai, and Bohan Zhuang. Ecoformer: Energy-saving atten- tion with linear complexity.Advances in Neural Informa- tion Processing Systems, 35:10295–10308,

    2022

  5. [23]

    Networks of spiking neu- rons: the third generation of neural network models.Neu- ral networks, 10(9):1659–1671,

    [Maass, 1997] Wolfgang Maass. Networks of spiking neu- rons: the third generation of neural network models.Neu- ral networks, 10(9):1659–1671,

    1997

  6. [25]

    Image super-resolution with non-local sparse attention

    [Meiet al., 2021 ] Yiqun Mei, Yuchen Fan, and Yuqian Zhou. Image super-resolution with non-local sparse attention. In Proceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition, pages 3517–3526,

    2021

  7. [26]

    Transformers for image recognition at scale.On- line: https://ai

    [Neil and Dirk, 2020] Houlsby Neil and Weissenborn Dirk. Transformers for image recognition at scale.On- line: https://ai. googleblog. com/2020/12/transformers- for-image-recognitionat. html,

    2020

  8. [27]

    X-linear attention networks for image captioning

    [Panet al., 2020 ] Yingwei Pan, Ting Yao, Yehao Li, and Tao Mei. X-linear attention networks for image captioning. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10971–10980,

    2020

  9. [28]

    Specaugment: A simple data augmentation method for automatic speech recognition.arXiv preprint arXiv:1904.08779,

    [Parket al., 2019 ] Daniel S Park, William Chan, Yu Zhang, Chung-Cheng Chiu, Barret Zoph, Ekin D Cubuk, and Quoc V Le. Specaugment: A simple data augmentation method for automatic speech recognition.arXiv preprint arXiv:1904.08779,

    2019 arXiv

  10. [30]

    Towards spike-based machine intelligence with neuromorphic computing.Nature, 575(7784):607–617,

    [Royet al., 2019 ] Kaushik Roy, Akhilesh Jaiswal, and Priyadarshini Panda. Towards spike-based machine intelligence with neuromorphic computing.Nature, 575(7784):607–617,

    2019

  11. [31]

    Spikingresformer: Bridging resnet and vision trans- former in spiking neural networks

    [Shiet al., 2024 ] Xinyu Shi, Zecheng Hao, and Zhaofei Yu. Spikingresformer: Bridging resnet and vision trans- former in spiking neural networks. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5610–5619,

    2024

  12. [33]

    Attention is all you need.Advances in neural information processing systems, 30,

    [Vaswaniet al., 2017 ] Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need.Advances in neural information processing systems, 30,

    2017

  13. [34]

    Pyramid vision transformer: A ver- satile backbone for dense prediction without convolutions

    [Wanget al., 2021 ] Wenhai Wang, Enze Xie, Xiang Li, Deng-Ping Fan, Kaitao Song, Ding Liang, Tong Lu, Ping Luo, and Ling Shao. Pyramid vision transformer: A ver- satile backbone for dense prediction without convolutions. InProceedings of the IEEE/CVF international conference o...

    2021

  14. [35]

    Generalisation of structural knowledge in the hippocampal-entorhinal system.Advances in neural information processing systems, 31,

    [Whittingtonet al., 2018 ] James Whittington, Timo- thy Muller, Shirely Mark, Caswell Barry, and Tim Behrens. Generalisation of structural knowledge in the hippocampal-entorhinal system.Advances in neural information processing systems, 31,

    2018

  15. [36]

    Attention spiking neural networks.arXiv preprint arXiv:2209.13929,

    [Yaoet al., 2022 ] Man Yao, Guangshe Zhao, Hengyu Zhang, Yifan Hu, Lei Deng, Yonghong Tian, Bo Xu, and Guoqi Li. Attention spiking neural networks.arXiv preprint arXiv:2209.13929,

    2022 arXiv

  16. [37]

    Metaformer is actually what you need for vision

    [Yuet al., 2022 ] Weihao Yu, Mi Luo, Pan Zhou, Chenyang Si, Yichen Zhou, Xinchao Wang, Jiashi Feng, and Shuicheng Yan. Metaformer is actually what you need for vision. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10819– 10829,

    2022

  17. [38]

    Spiking transformers for event-based single object track- ing

    [Zhanget al., 2022 ] Jiqing Zhang, Bo Dong, Haiwei Zhang, Jianchuan Ding, Felix Heide, Baocai Yin, and Xin Yang. Spiking transformers for event-based single object track- ing. InProceedings of the IEEE/CVF conference on Com- puter Vision and Pattern Recognition, pages 8801–8810,

    2022

  18. [39]

    Random erasing data aug- mentation

    [Zhonget al., 2020 ] Zhun Zhong, Liang Zheng, Guoliang Kang, Shaozi Li, and Yi Yang. Random erasing data aug- mentation. InProceedings of the AAAI conference on arti- ficial intelligence, volume 34, pages 13001–13008,

    2020

  19. [40]

    Spikformer: When spiking neural network meets transformer

    [Zhouet al., 2023 ] Zhaokun Zhou, Yuesheng Zhu, Chao He, Yaowei Wang, Shuicheng Yan, Yonghong Tian, and Li Yuan. Spikformer: When spiking neural network meets transformer. InICLR,

    2023

  20. [41]

    Spikformer v2: Join the high accuracy club on imagenet with an snn ticket.CoRR, 2024

    [Zhouet al., 2024b ] Zhaokun Zhou, Kaiwei Che, Wei Fang, Keyu Tian, Yuesheng Zhu, Shuicheng Yan, Yonghong Tian, and Li Yuan. Spikformer v2: Join the high accuracy club on imagenet with an snn ticket.CoRR, 2024

    2024

  21. [1984]

    Learning multiple layers of features from tiny im- ages

    [Krizhevskyet al., 2009 ] Alex Krizhevsky, Geoffrey Hinton, et al. Learning multiple layers of features from tiny im- ages

    2009

  22. [1997]

    Mobilevit: light-weight, general-purpose, and mobile-friendly vision transformer.arXiv preprint arXiv:2110.02178,

    [Mehta and Rastegari, 2021] Sachin Mehta and Mohammad Rastegari. Mobilevit: light-weight, general-purpose, and mobile-friendly vision transformer.arXiv preprint arXiv:2110.02178,

    2021 arXiv

  23. [2009]

    Bert: Pre-training of deep bidi- rectional transformers for language understanding.arXiv preprint arXiv:1810.04805,

    [Devlin, 2018] Jacob Devlin. Bert: Pre-training of deep bidi- rectional transformers for language understanding.arXiv preprint arXiv:1810.04805,

    2018 arXiv

  24. [2014]

    Deep networks with stochastic depth

    [Huanget al., 2016 ] Gao Huang, Yu Sun, Zhuang Liu, Daniel Sedra, and Kilian Q Weinberger. Deep networks with stochastic depth. InComputer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part IV 14, pages 646–

    2016

  25. [2016]

    Fact: Factor-tuning for lightweight adaptation on vision trans- former

    [Jie and Deng, 2023] Shibo Jie and Zhi-Hong Deng. Fact: Factor-tuning for lightweight adaptation on vision trans- former. InProceedings of the AAAI conference on artifi- cial intelligence, volume 37, pages 1060–1068,

    2023

  26. [2017]

    Segnet: A deep convolu- tional encoder-decoder architecture for image segmenta- tion.IEEE transactions on pattern analysis and machine intelligence, 39(12):2481–2495,

    [Badrinarayananet al., 2017 ] Vijay Badrinarayanan, Alex Kendall, and Roberto Cipolla. Segnet: A deep convolu- tional encoder-decoder architecture for image segmenta- tion.IEEE transactions on pattern analysis and machine intelligence, 39(12):2481–2495,

    2017

  27. [2018]

    Randaugment: Practical auto- mated data augmentation with a reduced search space

    [Cubuket al., 2020 ] Ekin D Cubuk, Barret Zoph, Jonathon Shlens, and Quoc V Le. Randaugment: Practical auto- mated data augmentation with a reduced search space. In Proceedings of the IEEE/CVF conference on computer vi- sion and pattern recognition workshops, pages 702–703,

    2020

  28. [2019]

    Towards artificial general intelligence with hybrid tianjic chip architecture.Nature, 572(7767):106–111,

    [Peiet al., 2019 ] Jing Pei, Lei Deng, Sen Song, Mingguo Zhao, Youhui Zhang, Shuang Wu, Guanrui Wang, Zhe Zou, Zhenzhi Wu, Wei He, et al. Towards artificial general intelligence with hybrid tianjic chip architecture.Nature, 572(7767):106–111,

    2019

  29. [2020]

    Imagenet: A large-scale hierarchical image database

    [Denget al., 2009 ] Jia Deng, Wei Dong, Richard Socher, Li- Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In2009 IEEE conference on computer vision and pattern recognition, pages 248–255. Ieee,

    2009

  30. [2021]

    Encoder-decoder with atrous separable convolution for se- mantic image segmentation

    [Chenet al., 2018 ] Liang-Chieh Chen, Yukun Zhu, George Papandreou, Florian Schroff, and Hartwig Adam. Encoder-decoder with atrous separable convolution for se- mantic image segmentation. InProceedings of the Euro- pean conference on computer vision (ECCV), pages 801– 818,

    2018

  31. [2022]

    1.1 computing’s energy problem (and what we can do about it)

    [Horowitz, 2014] Mark Horowitz. 1.1 computing’s energy problem (and what we can do about it). In2014 IEEE international solid-state circuits conference digest of tech- nical papers (ISSCC), pages 10–14. IEEE,

    2014

  32. [2023]

    The structure of im- ages.Biological cybernetics, 50(5):363–370,

    [Koenderink, 1984] Jan J Koenderink. The structure of im- ages.Biological cybernetics, 50(5):363–370,

    1984

  33. [2024]

    Training data-efficient image transformers & distillation through attention

    [Touvronet al., 2021 ] Hugo Touvron, Matthieu Cord, Matthijs Douze, Francisco Massa, Alexandre Sablay- rolles, and Herv ´e J ´egou. Training data-efficient image transformers & distillation through attention. InIn- ternational conference on machine learning, pages 10347–10357. PMLR,

    2021

Pith tools

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