Pith. sign in

REVIEW 4 major objections 6 minor 32 references

Context-aware thresholding sparsifies event-vision networks without sparsity-loss tuning.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

A learned input-adaptive threshold gates neuron activations in event-based neural networks, delivering near-SOTA object detection and optical flow with 32-68% fewer synaptic operations.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection Solid empirical method for sparsity in event-based vision; watch the recurrent threshold blind spot and missing error bars. the 4 major comments →

arxiv 2508.19806 v1 pith:XQ6J744Q submitted 2025-08-27 cs.CV cs.NE

Context-aware Sparse Spatiotemporal Learning for Event-based Vision

Shenqi Wang , Guangzhi Tang This is my paper
classification cs.CV cs.NE
keywords event-based visionneuromorphic computingactivation sparsitycontext-aware thresholdingobject detectionoptical flowsparse recurrent networksevent cameras
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

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 sets out to show that event-camera networks can become highly sparse without explicit sparsity regularization, simply by learning where to place the activation threshold. It introduces Context-aware Sparse Spatiotemporal Learning (CSSL), which computes a per-pixel threshold from the input feature distribution and zeros out every activation below that threshold. On event-based object detection, CSSL-SEED-256 reaches 46.4 mAP on the 1Mpx dataset with 2.80 GSOp, while the SEED-256 baseline it builds on reaches 44.9 mAP with 3.83 GSOp. On optical flow, CSSL-EV-FlowNet reaches 2.38 average endpoint error at 10.61% neuron density, compared with 2.88 and 16.90% for the recurrent baseline. The practical significance is that robot perception with event cameras could run at a fraction of the compute, without fragile sparsity-loss hyperparameters.

Core claim

CSSL's central claim is that dynamic context-aware thresholding can replace both fixed ReLU thresholds and explicit sparsity regularization in event-based vision. At each time step a small convolution reads the input feature map and emits a per-pixel threshold v_th = sigma(W_v x + b_v) constrained to [0,1]; a Heaviside step then produces a binary mask that zeroes every output activation below that threshold. Because the threshold tracks the input distribution, the network concentrates computation on informative locations and naturally keeps activation density low. The authors show the mechanism generalizes across convolutional, residual, and recurrent blocks and across object detection and o

What carries the argument

The central object is the context-aware thresholding gate: a per-pixel threshold v_th = sigma(W_v x + b_v) produced by a convolution over the input, followed by a Heaviside mask s = H(y~ - v_th). It does the sparsification work that sparsity loss terms normally do. In recurrent blocks the threshold is computed from the previous sparse hidden state y(t-1) rather than the current input, and a soft-reset subtracts the threshold after the unit fires; this regulates spatiotemporal sparsity while keeping the dense input convolution cheap.

Load-bearing premise

The load-bearing premise is that a learned per-pixel threshold, hard-zeroing every activation below it, reliably separates informative from redundant signals in every layer and every task, and that in recurrent layers the threshold can be read from the previous hidden state alone.

What would settle it

Run a trained CSSL-SEED on a low-contrast, small-object subset of an event-detection benchmark and compare it with the same architecture with all learned thresholds forced to zero at inference. If the dense version's mAP is materially higher, the hard threshold is discarding task-critical features; if it is not, the masking is information-preserving on that distribution.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Sparsity becomes an architectural property rather than a training objective, so practitioners can drop the beta_sparse hyperparameter and the two-stage tuning it requires.
  • CSSL modules can replace ordinary convolutions, residual blocks, and recurrent cells in existing architectures; the paper demonstrates this on detection and optical flow with consistent efficiency gains.
  • The framework works across multiple recurrent cells (MGU, GRU, MinimalRNN), with MGU and GRU nearly matching each other and MinimalRNN trading some accuracy for fewer parameters and operations.
  • On neuromorphic processors that charge per synaptic operation or per active neuron, the reduced activation density should translate directly into lower energy and faster inference, which the paper positions as the motivation for the framework.

Where Pith is reading between the lines

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

  • Because the recurrent threshold in Eq. (2b) sees only the previous hidden state and not the current input, abrupt scene changes that generate many fresh events while the hidden state is stale are an untested stress case; the paper does not analyze this failure mode.
  • The masking operator is a hard zero-out, so measuring how much task-relevant information actually sits below the learned thresholds (e.g., by ablating the mask at inference) would reveal where the efficiency-accuracy trade-off truly lives.
  • The same input-conditional gating idea could be applied to attention-based event backbones, where sparsity would prune tokens or channels rather than pixels, but the paper only demonstrates convolutional and recurrent modules.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes Context-aware Sparse Spatiotemporal Learning (CSSL), a training/inference mechanism that adds per-pixel, input-dependent threshold convolutions to standard convolutional and convolutional-recurrent units. The threshold is produced by a sigmoid-bounded convolution (Eq. 1) and applied through a Heaviside gate, giving sparse activations without an explicit sparsity loss. The method is instantiated in two architectures: CSSL-SEED for event-based object detection and CSSL-EV-FlowNet for optical flow. On the 1Mpx and Gen1 detection benchmarks it reports mAP close to or above the SEED baseline at lower GSOp, and on MVSEC it reports lower AEE and lower activation density than RNN-EV-FlowNet. The authors argue that CSSL removes the need to tune sparsity-loss weights and is suitable for neuromorphic deployment.

Significance. If the reported results hold, CSSL is a useful and simple technique for obtaining high activation sparsity in event-based vision networks, and the code release supports reproducibility. The comparison against an explicit sparsity-loss baseline (Table IV) is informative and shows that the learned thresholds attain a better accuracy/efficiency trade-off than L1 sparsification. However, the strength of the evidence is limited by single-run experiments with small margins, by the unclear cost accounting of the auxiliary threshold convolutions, and by the fact that the recurrent threshold in Eq. (2b) is not actually conditioned on the current input. The core idea is plausible and the empirical results are encouraging, but the load-bearing efficiency and generality claims need stronger support before publication.

major comments (4)
  1. [§III-B, Tables II and III] All reported detection and flow results appear to come from a single training run. The central claim of 'comparable or superior performance' rests on small differences, e.g., CSSL-SEED-256 46.4 vs. SEED-256 44.9 mAP on 1Mpx, and 46.3 vs. 45.3 on Gen1. Without error bars, multiple seeds, or significance tests it is impossible to know whether these gaps are systematic or training noise. Please report mean and std over at least 3 runs, or, if that is computationally infeasible, provide a justification and at least repeat the most critical comparisons (e.g., SEED-256 vs. CSSL-SEED-256 and RNN-EV-FlowNet vs. CSSL-EV-FlowNet).
  2. [§II-D, Eq. (2b)] The recurrent threshold v_th(t) depends only on the previous sparse hidden state y(t-1), not on the current input x(t). For a pixel that was inactive at t-1, v_th(t) reduces to sigma(b_v), a learned constant. If this constant is large, current-input-driven activations in c(t) are zeroed by the Heaviside gate in Eq. (2e), so newly appearing features can be suppressed before they propagate. The paper offers no analysis of how often this destructive masking occurs or whether it correlates with misses on the benchmark tasks. Since the abstract claims thresholds are based on the input distribution, this is a gap. Please add an ablation where v_th also depends on x(t) (e.g., concatenating x(t) with y(t-1) before the threshold convolution) and/or analyze failure cases to show that newly appearing objects are not systematically thresholded out.
  3. [§II-B/Eq. (1), Table II] The reported GSOp numbers do not clearly state whether they include the extra threshold-generating convolution W_v in Eq. (1) and the threshold convolution W_v in Eq. (2b). These are additional synaptic operations, so the efficiency comparison depends on this accounting. If these auxiliary convolutions are excluded from GSOp, the CSSL efficiency advantage is overstated. In addition, the sentence in Section II-D claiming that 'the dense tensor has a negligible impact on the total number of synaptic operations' is not quantified; the multiplications Wxf*x(t) and Wxh*x(t) in Eq. (2) still contribute to the total. Please provide an explicit cost model or measured operation counts that include all auxiliary convolutions.
  4. [§II-C, Fig. 2(b)] The context-aware residual block is described only in prose and the figure; no equations are given for how the second convolution is split into threshold and dense components, how the post-accumulation threshold is computed, or how the sparse output is formed. This is not only a reproducibility issue but also makes it hard to verify that the residual-block implementation in Table II matches the described mechanism. Please add the corresponding equations or pseudocode.
minor comments (6)
  1. [Abstract] Typo: 'struggle to match of performance' should be 'struggle to match the performance'.
  2. [Throughout] The manuscript frequently uses 'we proposed' where 'we propose' is intended. Please correct.
  3. [Table V vs Table II] Table V lists 'SEED without sparsity loss' with mAP 45.0, while Table II reports SEED-256 with mAP 44.9. Please clarify whether these are the same model or different configurations.
  4. [§III-D] The claim that CSSL 'eliminates the need for manually tuned sensitive sparsity hyperparameters' is supported only for the L1 sparsity weight. The surrogate gradient scale and the sigmoid threshold range are also hyperparameters, but their sensitivity is not analyzed. A short sensitivity study would strengthen the claim.
  5. [§II-D/§III-C] The soft-reset mechanism in Eq. (3) is introduced as beneficial for forgetting and training stability, but no ablation isolates its contribution. Please report a comparison with and without the soft reset.
  6. [§III-A] The relationship between the optical-flow training setup and the MVSEC evaluation protocol is taken directly from [7]; please state explicitly whether the reported RNN-EV-FlowNet numbers are copied from [7] or re-run, and if re-run, whether the same code/config was used.

Circularity Check

0 steps flagged

No significant circularity: CSSL's thresholds are learned end-to-end and evaluated on held-out benchmarks; self-cited baselines are comparators, not derivation inputs.

full rationale

The paper's central claim is empirical. The context-aware threshold (Eq. 1: v_th = σ(W_v x + b_v); mask s = H(ỹ - v_th)) and the recurrent threshold (Eq. 2b: v_th = σ(W_v y(t-1) + b_v)) are learned parameters optimized by task loss through surrogate gradients; no reported mAP, GSOp, AEE, or activation density is a fitted parameter renamed as a prediction. Performance is measured on held-out splits of public datasets (1Mpx, Gen1, UZH-FPV, MVSEC), against both same-group baselines (SEED [8], RNN-EV-FlowNet [7]) and independent methods (RVT, ASTMNet, SpikeYOLO, etc.). The self-citations are used as architecture/baseline references, not as an unverified uniqueness theorem or as the justificatory source of the sparse-activation result. The wording that the threshold 'selectively retains informative features' is definitional (what passes the threshold is called informative), but this does not carry the paper's quantitative claims, which rest on measured mAP/AEE/density. The recurrent threshold's dependence only on y(t-1) is a potential robustness limitation for newly appearing input features, but that is a modeling/correctness concern, not a circular reduction: the reported improvements do not reduce to the threshold formula by construction. Therefore no circular step is present.

Axiom & Free-Parameter Ledger

2 free parameters · 4 axioms · 1 invented entities

CSSL is an empirical method rather than a derivation. Its central claim rests on several hand-chosen architectural assumptions: the fixed [0,1] range for the threshold, the inherited surrogate gradient, the event-based operation cost model, and the soft-reset rule. None of these are independently validated outside the paper's own experiments.

free parameters (2)
  • Threshold sigmoid range [0,1] = [0,1] (fixed)
    v_th = sigma(...) constrains thresholds to [0,1] (Eq. 1); this hand-chosen range determines which activations are retained and is not derived from data or task statistics.
  • Surrogate gradient scale = inherited from [12]
    The paper uses the surrogate gradient method of Subramoney et al. [12] to train through the Heaviside step, and does not specify or tune the surrogate's scale or shape, treating it as an external fixed choice.
axioms (4)
  • domain assumption The Heaviside step function H with a surrogate gradient can be trained end-to-end in this architecture.
    Used in Eq. 1 for s(t) and Eq. 2e; the paper relies on [12] without local validation that the surrogate does not harm gradient flow in deep conv-recurrent nets.
  • domain assumption The event-based convolution synaptic-operation cost model accurately reflects computational cost.
    GSOp comparisons in Tables II and III assume only operations on active spikes are counted, per [8,11], and that the extra threshold convolution has negligible cost; no hardware energy measurement is provided.
  • ad hoc to paper A sigmoid-bounded threshold is a sufficient relevance filter for all layers.
    No theoretical justification is given that pixel-wise thresholding with a [0,1] range preserves informative activations; the claim is purely empirical (Section II-B).
  • ad hoc to paper The soft-reset mechanism (Eq. 3) helps forgetting and stabilizes training.
    Introduced 'inspired by spiking neurons' without ablations or analysis of its effect on gradient flow or state dynamics.
invented entities (1)
  • Auxiliary hidden state c(t) no independent evidence
    purpose: Dense memory that can be thresholded to produce a sparse output y(t) in conv-recurrent units (Eq. 2d-2f).
    A purely computational construct introduced by the paper with no external measurement or falsifiable handle; it is an architectural variable.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Context-aware Sparse Spatiotemporal Learning for Event-based Vision." pith.science (2026). https://pith.science/paper/XQ6J744Q

@misc{pith2026250819806,
  author       = {Pith},
  title        = {Pith review of: Context-aware Sparse Spatiotemporal Learning for Event-based Vision},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XQ6J744Q}},
  note         = {Machine review of arXiv:2508.19806}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Event-based camera has emerged as a promising paradigm for robot perception, offering advantages with high temporal resolution, high dynamic range, and robustness to motion blur. However, existing deep learning-based event processing methods often fail to fully leverage the sparse nature of event data, complicating their integration into resource-constrained edge applications. While neuromorphic computing provides an energy-efficient alternative, spiking neural networks struggle to match of performance of state-of-the-art models in complex event-based vision tasks, like object detection and optical flow. Moreover, achieving high activation sparsity in neural networks is still difficult and often demands careful manual tuning of sparsity-inducing loss terms. Here, we propose Context-aware Sparse Spatiotemporal Learning (CSSL), a novel framework that introduces context-aware thresholding to dynamically regulate neuron activations based on the input distribution, naturally reducing activation density without explicit sparsity constraints. Applied to event-based object detection and optical flow estimation, CSSL achieves comparable or superior performance to state-of-the-art methods while maintaining extremely high neuronal sparsity. Our experimental results highlight CSSL's crucial role in enabling efficient event-based vision for neuromorphic processing.

Figures

Figures reproduced from arXiv: 2508.19806 by Guangzhi Tang, Shenqi Wang.

Figure 1
Figure 1. Figure 1: An overview of the proposed Context-aware Sparse Spatiotemporal [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Illustration of the context-aware thresholding process. The input feature is processed through a context-aware convolution that generates an [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

discussion (0)

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

Reference graph

Works this paper leans on

32 extracted references · 27 canonical work pages · 2 internal anchors

  1. [1]

    Event- based vision: A survey,

    G. Gallego, T. Delbr ¨uck, G. Orchard, C. Bartolozzi, B. Taba, A. Censi, S. Leutenegger, A. J. Davison, J. Conradt, K. Daniilidis, et al., “Event- based vision: A survey,” IEEE transactions on pattern analysis and machine intelligence, vol. 44, no. 1, pp. 154–180, 2020

    2020

  2. [2]

    Low latency automotive vision with event cameras,

    D. Gehrig and D. Scaramuzza, “Low latency automotive vision with event cameras,” 2024

    work page 2024

  3. [3]

    Monocular Event-Based Vision for Obstacle Avoidance with a Quadrotor

    A. Bhattacharya, M. Cannici, N. Rao, Y . Tao, V . Kumar, N. Matni, and D. Scaramuzza, “Monocular event-based vision for obstacle avoidance with a quadrotor,” 2024. [Online]. Available: https://arxiv.org/abs/2411.03303

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  4. [4]

    Effective sensor fusion with event-based sensors and deep network architectures,

    D. Neil and S.-C. Liu, “Effective sensor fusion with event-based sensors and deep network architectures,” in 2016 IEEE International Symposium on Circuits and Systems (ISCAS) , 2016, pp. 2282–2285

    work page 2016

  5. [5]

    The neurobench framework for benchmarking neuromorphic comput- ing algorithms and systems,

    J. Yik, K. Van den Berghe, D. den Blanken, Y . Bouhadjar, M. Fabre, P. Hueber, W. Ke, M. A. Khoei, D. Kleyko, N. Pacik-Nelson, et al. , “The neurobench framework for benchmarking neuromorphic comput- ing algorithms and systems,” Nature communications, vol. 16, no. 1, p. 1545, 2025

    work page 2025

  6. [6]

    Slayer: Spike layer error reassignment in time,

    S. B. Shrestha and G. Orchard, “Slayer: Spike layer error reassignment in time,” in Advances in Neural Information Processing Systems , S. Bengio, H. Wallach, H. Larochelle, K. Grauman, N. Cesa-Bianchi, and R. Garnett, Eds., vol. 31. Curran Associates, Inc., 2018. [Online]. Available: https://proceedings.neurips.cc/paper files/paper/ 2018/file/82f2b308c3b...

    work page 2018

  7. [7]

    Event-based optical flow on neuromorphic processor: Ann vs. snn comparison based on activation sparsification,

    Y . Xu, G. Tang, A. Yousefzadeh, G. C. de Croon, and M. Sifalakis, “Event-based optical flow on neuromorphic processor: Ann vs. snn comparison based on activation sparsification,” Neural Networks, vol. 188, p. 107447, 2025

    work page 2025

  8. [8]

    Sparse convolutional recurrent learning for ef- ficient event-based neuromorphic object detection,

    S. Wang, Y . Xu, A. Yousefzadeh, S. Eissa, H. Corporaal, F. Cor- radi, and G. Tang, “Sparse convolutional recurrent learning for ef- ficient event-based neuromorphic object detection,” arXiv preprint arXiv:2506.13440, 2025

    Pith/arXiv arXiv 2025

  9. [9]

    Conditional computation in neural networks for faster models,

    E. Bengio, P.-L. Bacon, J. Pineau, and D. Precup, “Conditional computation in neural networks for faster models,” arXiv preprint arXiv:1511.06297, 2015

    Pith/arXiv arXiv 2015

  10. [10]

    Deja vu: Contextual sparsity for efficient LLMs at inference time,

    Z. Liu, J. Wang, T. Dao, T. Zhou, B. Yuan, Z. Song, A. Shrivastava, C. Zhang, Y . Tian, C. Re, and B. Chen, “Deja vu: Contextual sparsity for efficient LLMs at inference time,” in Proceedings of the 40th International Conference on Machine Learning , ser. Proceedings of Machine Learning Research, A. Krause, E. Brunskill, K. Cho, B. Engelhardt, S. Sabato, ...

    work page 2023

  11. [11]

    Optimizing event-based neural networks on digital neuromorphic architecture: a comprehensive design space exploration,

    Y . Xu, K. Shidqi, G.-J. van Schaik, R. Bilgic, A. Dobrita, S. Wang, R. Meijer, P. Nembhani, C. Arjmand, P. Martinello, A. Gebregiorgis, S. Hamdioui, P. Detterer, S. Traferro, M. Konijnenburg, K. Vadivel, M. Sifalakis, G. Tang, and A. Yousefzadeh, “Optimizing event-based neural networks on digital neuromorphic architecture: a comprehensive design space ex...

    work page arXiv 2024

  12. [12]

    Efficient recurrent architectures through activity sparsity and sparse back-propagation through time,

    A. Subramoney, K. K. Nazeer, M. Sch ¨one, C. Mayr, and D. Kappel, “Efficient recurrent architectures through activity sparsity and sparse back-propagation through time,” in The Eleventh International Con- ference on Learning Representations , 2022

    work page 2022

  13. [13]

    Minimal gated unit for recurrent neural networks,

    G.-B. Zhou, J. Wu, C.-L. Zhang, and Z.-H. Zhou, “Minimal gated unit for recurrent neural networks,” International Journal of Automation and Computing, vol. 13, no. 3, pp. 226–234, 2016

    work page 2016

  14. [14]

    Reducing information loss for spiking neural networks,

    Y . Guo, Y . Chen, L. Zhang, Y . Wang, X. Liu, X. Tong, Y . Ou, X. Huang, and Z. Ma, “Reducing information loss for spiking neural networks,” in Computer Vision – ECCV 2022 , S. Avidan, G. Brostow, M. Ciss ´e, G. M. Farinella, and T. Hassner, Eds. Cham: Springer Nature Switzerland, 2022, pp. 36–52

    work page 2022

  15. [15]

    Learning to detect objects with a 1 megapixel event camera,

    E. Perot, P. De Tournemire, D. Nitti, J. Masci, and A. Sironi, “Learning to detect objects with a 1 megapixel event camera,”Advances in Neural Information Processing Systems , vol. 33, pp. 16 639–16 652, 2020

    work page 2020

  16. [16]

    A large scale event-based detection dataset for automotive,

    P. De Tournemire, D. Nitti, E. Perot, D. Migliore, and A. Sironi, “A large scale event-based detection dataset for automotive,” arXiv preprint arXiv:2001.08499, 2020

    Pith/arXiv arXiv 2001

  17. [17]

    Microsoft coco: Common objects in context,

    T.-Y . Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Doll ´ar, and C. L. Zitnick, “Microsoft coco: Common objects in context,” in Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part V 13 . Springer, 2014, pp. 740–755

    2014

  18. [18]

    Ssd: Single shot multibox detector,

    W. Liu, D. Anguelov, D. Erhan, C. Szegedy, S. Reed, C.-Y . Fu, and A. C. Berg, “Ssd: Single shot multibox detector,” in Computer Vision– ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part I 14 . Springer, 2016, pp. 21–37

    work page 2016

  19. [19]

    Are we ready for autonomous drone racing? the uzh-fpv drone racing dataset,

    J. Delmerico, T. Cieslewski, H. Rebecq, M. Faessler, and D. Scara- muzza, “Are we ready for autonomous drone racing? the uzh-fpv drone racing dataset,” in 2019 International Conference on Robotics and Automation (ICRA). IEEE, 2019, pp. 6713–6719

    work page 2019

  20. [20]

    The multivehicle stereo event camera dataset: An event camera dataset for 3d perception,

    A. Z. Zhu, D. Thakur, T. ¨Ozaslan, B. Pfrommer, V . Kumar, and K. Daniilidis, “The multivehicle stereo event camera dataset: An event camera dataset for 3d perception,” IEEE Robotics and Automation Letters, vol. 3, no. 3, pp. 2032–2039, 2018

    work page 2032

  21. [21]

    Recurrent vision transformers for object detection with event cameras,

    M. Gehrig and D. Scaramuzza, “Recurrent vision transformers for object detection with event cameras,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 13 884–13 893

    work page 2023

  22. [22]

    Asynchronous spatio-temporal memory network for continuous event-based object detection,

    J. Li, J. Li, L. Zhu, X. Xiang, T. Huang, and Y . Tian, “Asynchronous spatio-temporal memory network for continuous event-based object detection,” IEEE Transactions on Image Processing, vol. 31, pp. 2975– 2987, 2022

    work page 2022

  23. [23]

    Object detection with spiking neural networks on automotive event data,

    L. Cordone, B. Miramond, and P. Thierion, “Object detection with spiking neural networks on automotive event data,” in 2022 Interna- tional Joint Conference on Neural Networks (IJCNN) . IEEE, 2022, pp. 1–8

    work page 2022

  24. [24]

    Integer-valued training and spike-driven inference spiking neural network for high- performance and energy-efficient object detection,

    X. Luo, M. Yao, Y . Chou, B. Xu, and G. Li, “Integer-valued training and spike-driven inference spiking neural network for high- performance and energy-efficient object detection,” inComputer Vision – ECCV 2024 , A. Leonardis, E. Ricci, S. Roth, O. Russakovsky, T. Sattler, and G. Varol, Eds. Cham: Springer Nature Switzerland, 2025, pp. 253–272

    work page 2024

  25. [25]

    Self-supervised learning of event-based optical flow with spiking neural networks,

    J. Hagenaars, F. Paredes-Vall ´es, and G. De Croon, “Self-supervised learning of event-based optical flow with spiking neural networks,” Advances in Neural Information Processing Systems , vol. 34, pp. 7167–7179, 2021

    work page 2021

  26. [26]

    MinimalRNN: Toward More Interpretable and Trainable Recurrent Neural Networks

    M. Chen, “Minimalrnn: Toward more interpretable and trainable recurrent neural networks,” arXiv preprint arXiv:1711.06788 , 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  27. [27]

    Learning phrase representations using rnn encoder-decoder for statistical machine translation,

    K. Cho, B. van Merrienboer, C. Gulcehre, D. Bahdanau, F. Bougares, H. Schwenk, and Y . Bengio, “Learning phrase representations using rnn encoder-decoder for statistical machine translation,” 2014

    work page 2014

  28. [28]

    Towards artificial general intelligence with hybrid tianjic chip architecture,

    J. Pei, L. Deng, S. Song, M. Zhao, Y . Zhang, S. Wu, G. Wang, Z. Zou, Z. Wu, W. He, et al. , “Towards artificial general intelligence with hybrid tianjic chip architecture,” Nature, vol. 572, no. 7767, pp. 106– 111, 2019

    work page 2019

  29. [29]

    Memory- efficient deep learning on a spinnaker 2 prototype,

    C. Liu, G. Bellec, B. V ogginger, D. Kappel, J. Partzsch, F. Neum ¨arker, S. H ¨oppner, W. Maass, S. B. Furber, R. Legenstein, et al., “Memory- efficient deep learning on a spinnaker 2 prototype,” Frontiers in neuroscience, vol. 12, p. 840, 2018

    work page 2018

  30. [30]

    Opportunities for neuromorphic computing algorithms and applications,

    C. D. Schuman, S. R. Kulkarni, M. Parsa, J. P. Mitchell, B. Kay, et al. , “Opportunities for neuromorphic computing algorithms and applications,” Nature Computational Science , vol. 2, no. 1, pp. 10– 19, 2022

    work page 2022

  31. [31]

    From chaos comes order: Ordering event representations for object recognition and detection,

    N. Zubi ´c, D. Gehrig, M. Gehrig, and D. Scaramuzza, “From chaos comes order: Ordering event representations for object recognition and detection,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2023, pp. 12 846–12 856

    work page 2023

  32. [32]

    Better and faster: Adaptive event conversion for event-based object detection,

    Y . Peng, Y . Zhang, P. Xiao, X. Sun, and F. Wu, “Better and faster: Adaptive event conversion for event-based object detection,” in Pro- ceedings of the AAAI Conference on Artificial Intelligence , vol. 37, no. 2, 2023, pp. 2056–2064

    work page 2023

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.