Pith. sign in

REVIEW 5 major objections 6 minor 56 references

Latency Coding for Efficient and Low-Latency Deep Spiking Neural Networks

T0 review · 5 major / 6 minor · reviewed 2026-08-02 · deepseek-v4-flash

Pith's one-line read The paper claims that time-to-first-spike (TTFS) spiking networks can be trained with backpropagation through time, reaching state-of-the-art accuracy among TTFS methods at ultra-low latency by allowing hidden neurons to fire multiple times

desk verdict A genuine training recipe for latency-coded SNNs with strong accuracy claims, but the headline latency comparison to prior TTFS methods is not apples-to-apples and needs a common timing convention. read the letter →

arxiv 2603.23206 v2 pith:EFOYZZBO submitted 2026-03-24 cs.NE

classification cs.NE
keywords spikingneuralnetworkstime-to-first-spikecodinglatencybackpropagationthroughtimetemporallow-latencyinferenceneuromorphiccomputingenergyefficiency
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 aims to make time-to-first-spike (TTFS) coding practical for deep spiking neural networks by training latency-coded SNNs with backpropagation through time (BPTT). It claims that by relaxing the strict single-spike constraint in hidden layers, using a feature-extraction encoder, and weighting loss by the model's confidence, TTFS networks can reach state-of-the-art accuracy among temporal-coding methods while completing inference in about one to four timesteps. If true, latency coding becomes a hardware-friendly alternative to rate coding for fast, energy-efficient neuromorphic inference. The paper also reports improved robustness to input corruptions relative to rate-coded SNNs.

What carries the argument

The machinery is the iterative LIF neuron with soft reset (Eq. 2-4), combined with three components: a latency encoding module that maps extracted features to spike times via t_s(x)=ceil((1-x)T) and passes gradients via a straight-through estimator; a decoding rule that selects the class by earliest output spike and breaks ties using membrane potential (Eq. 11); and the temporal adaptive decision (TAD) loss, which weights cross-entropy at each timestep by a temperature-smoothed inverse-entropy confidence (Eq. 12-14). Together, these allow BPTT to train a network whose decision is made from first-spike timing while hidden layers retain multi-spike gradient flow.

What would settle it

Implement the proposed latency-coded network and a prior TTFS baseline from the paper's comparison table on the same neuromorphic simulator with a strictly layer-wise clock (each layer takes one timestep to propagate), and measure wall-clock time to first output spike on CIFAR-10. If the latency-coded network does not remain at least an order of magnitude faster than the baseline when both use identical time units and communication protocols, the central latency claim fails.

Watch

Extended reading notes

Core claim

The central claim is that latency-coded SNNs—where information is carried by the timing of the first output spike—can be trained end-to-end with BPTT once three design choices are made. A latency encoding module extracts features and encodes those features, not raw pixels, into spike times using a straight-through estimator. Hidden-layer neurons are allowed to fire multiple times, while the output layer still decides based on the earliest spike, with membrane potential used to break ties. A temporal adaptive decision (TAD) loss weights per-timestep cross-entropy by the network's confidence, pushing easy samples to fire early and hard samples to integrate longer. The paper reports 93.60% on C

Load-bearing premise

The load-bearing assumption is that the reported inference time—measured as the first output spike in a synchronous layer-wise simulation where a spike can propagate through all layers in a single global timestep—is directly comparable to the per-layer timestep counts reported for prior TTFS methods (for example, thousands of timesteps per layer). If these metrics are not commensurate, the claimed two-orders-of-magnitude latency reduction is partly a metering artifact rather

Editorial extensions

If this is right

  • Latency-coded SNNs can run inference in 1-4 timesteps on standard vision benchmarks, making real-time neuromorphic deployment feasible.
  • BPTT becomes a viable training paradigm for temporal coding, enabling deeper TTFS networks without conversion or event-driven learning rules.
  • Relaxing the single-spike constraint in hidden layers while keeping first-spike output decisions resolves gradient vanishing without increasing decision latency.
  • The reported robustness advantage suggests temporal coding may be preferable to rate coding under input corruptions, not just for speed.
  • Energy estimates indicate latency-coded SNNs consume a small fraction of the energy of ANN counterparts and less than rate-coded SNNs.

Reading between the lines

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

  • If the inference-latency metric is genuinely commensurate with prior work, the framework could be adapted to asynchronous event-driven hardware, where the wall-clock latency would be set by the critical path through layers rather than the number of global timesteps.
  • The TAD loss's confidence-based weighting scheme is a general principle that could transfer to other early-exit or adaptive-computation architectures, not only spiking networks.
  • The low temporal similarity of latency-coded representations suggests a testable hypothesis: the robustness gain comes from decorrelated per-timestep features, which could be verified by ablating the TAD loss and measuring both robustness and temporal similarity.
  • A direct extension would be to measure energy on actual neuromorphic chips, since the theoretical energy model assumes zero static-energy scaling with latency; the claimed advantage may shrink on hardware with high static power.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 6 minor

Summary. The paper proposes a framework for directly training deep time-to-first-spike (TTFS) / latency-coded spiking neural networks with BPTT and surrogate gradients. Three components are introduced: a latency encoding (LE) module with a straight-through estimator (Eqs. 5-9), relaxation of the single-spike constraint in hidden layers with a membrane-potential-based decoding rule in the output layer (Eqs. 10-11), and a temporal adaptive decision (TAD) loss that reweights cross-entropy over timesteps by confidence (Eqs. 12-14). The experiments report strong accuracy at 1-4 inference timesteps on CIFAR-10, CIFAR-100, Tiny-ImageNet, and CIFAR10-DVS, along with theoretical energy estimates and a robustness comparison against rate-coded SNNs. The central claim is that latency-coded SNNs can be trained efficiently with BPTT and achieve state-of-the-art accuracy among TTFS-coded SNNs with ultra-low latency and high energy efficiency.

Significance. If the reported latency metric is commensurate with previous TTFS work, the result is significant: it would make BPTT-trained temporal-coding SNNs competitive with rate-coded SNNs at a fraction of the inference latency, and the proposed LE/TAD components are clearly characterized by ablations and time-scalability experiments (Fig. 6). The paper also offers a useful temporal-similarity analysis (Sec. IV-D). However, the headline comparisons currently mix architectures, use an inconsistent time-counting protocol, and lack uncertainty estimates; the claims therefore need to be re-baselined before the result can be accepted as stated.

major comments (5)
  1. [Sec. III-A, Eq. (2), Table I] The central latency comparison is not measured under a common protocol. Eq. (2) uses S^{l-1}[t] in the update of U^l[t] in the same global timestep, so with the synchronous layer-wise update a signal can travel from input to output within one timestep. The baselines in Table I count time differently: [34] is listed as '4096 per layer', and Sec. I states that prior TTFS methods need 'multiplication of quantized steps and the number of layers'. Consequently, the reported 1.00-4.00 inference timesteps and the 'two orders of magnitude' speedup are not directly comparable to the per-layer timestep counts of DTA-TTFS, T2FSNN, TSC-SNN, etc. Please report the comparison under a common timing convention (e.g., a delay-aware or layer-serial simulation) and restate the latency/energy claims accordingly.
  2. [Sec. IV-A, Table I] The accuracy comparisons are confounded by architecture and training setup. All TTFS baselines in Table I are VGG-16, while the proposed results use VGG-11, VGG-16, and SEW-ResNet-18; on CIFAR-10 the proposed best accuracy (93.64 with SEW-ResNet-18) is below the 93.69 of Stanojevic et al. (VGG-16), so the 'state-of-the-art accuracy' claim is not supported. On CIFAR-100 the gain over the best baseline (72.24 vs 74.97) is large, but the architecture differs (VGG-11 vs VGG-16). No error bars or seed-to-seed variability are reported, and augmentation recipes differ among datasets. Please provide same-architecture comparisons and multi-seed statistics for the main results.
  3. [Sec. II-A, Sec. III-A, Eq. (3)] The BPTT/surrogate-gradient training is not fully specified. The firing function H(·) in Eq. (3) is treated as a Heaviside step, and an STE is defined only for the latency encoder (Eq. 9). No surrogate gradient is given for the spike function used in backpropagation through the hidden layers. Since surrogate choice and slope can substantially change accuracy and latency in SNNs, please state the exact surrogate gradient (including its functional form and hyperparameters) and, ideally, ablate it. This is required for reproducibility of the central method.
  4. [Sec. IV-B, Eq. (17), Table III] The normalized-energy comparison inherits the timing-protocol problem. Eq. (17) multiplies a static-energy coefficient by 'Timesteps'; for the proposed model this is the zero-delay global timestep count, whereas for baselines it is the per-layer timestep count from Table I. The '0.366x/0.492x' energy values therefore understate the proposed model's energy if a hardware-compatible per-layer delay is used. Please recompute Table III with a common time granularity, or explicitly justify a hardware mapping in which same-timestep propagation across all layers is realizable.
  5. [Sec. IV-C, Table IV] The robustness advantage is supported by only a single paired comparison with no statistical significance. On CIFAR-10-C, the mCE gap is 0.8 points (33.9 vs 34.7) and the latency model has a higher clean error (6.4 vs 5.4); on CIFAR-100-C the gap is larger (50.8 vs 56.9). Without multiple seeds or confidence intervals, the robustness claim in the abstract is premature. Please provide variance estimates or down-rank the claim.
minor comments (6)
  1. [Throughout] Several typos: 'backpropagation throuh time' (Sec. I), 'serveritis' (Table IV caption), 'comparisom' (Sec. IV-E2), 'Ploted' (Sec. IV-C). Please proofread.
  2. [Eq. (8)] Since F is defined via Sigmoid(·) in (0,1), the latency mapping is well-defined only for x in (0,1); please state the treatment of boundary values (or note that sigmoid never reaches 0 or 1).
  3. [Sec. III-C, Eqs. (12)-(14)] O[t], the pre-synaptic current of the output layer, is used in the TAD loss but not defined in the model description. Please define it and clarify its relationship to the logits used in the decoding rule.
  4. [Eq. (10)] The variable T is reused for both the maximum training timestep and the earliest spike time; rename one to avoid ambiguity.
  5. [Table III] The column header is garbled ('Neural Time ... Spikes (10^4)'); please format clearly and state exactly how 'Normalized Energy' is computed from Timesteps and Spikes.
  6. [Reproducibility] The manuscript would benefit from a reproducibility statement and, ideally, code/checkpoints, as no such information is currently provided.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the training framework and reported latency/accuracy/energy results are measured outcomes, not consequences of a fitted premise or a self-citation chain.

full rationale

I find no circular step. The paper proposes an engineering/training framework for latency-coded SNNs: a latency encoding (LE) module, relaxed single-spike constraint, and a temporal adaptive decision (TAD) loss, all trained with BPTT. The reported accuracies, inference timesteps, and energy estimates are measured on held-out test sets rather than derived from the model's assumptions. The TAD loss explicitly optimizes early confident firing, so the resulting low latency is an outcome being optimized by the loss, not a fitted parameter renamed as a prediction. The paper's own equation ts(x)=ceil((1-x)T) is a standard intensity-to-latency encoding choice, not a hidden restatement of the conclusion. The robustness analysis in Sec. IV-D is post-hoc interpretation of measured temporal similarity, not a derivation from the framework. Self-citations appear (e.g., SEW-ResNet [37] and some energy references), but they are not load-bearing: the architecture is an externally established public model, and no central claim is justified solely by a same-author citation. The only substantive concern is the comparability of latency counts in Table I: Eq. (2) permits a spike from layer l-1 at time t to contribute to layer l in the same global timestep, so an 11-layer network can emit an output spike at t=1, whereas some baselines are quoted as 4096 timesteps per layer. That is a measurement-protocol or external-validity concern, not circularity: the proposed model's own result is not equivalent to its input by construction. Therefore the circularity score is 0.

Assumptions & free parameters 4 free parameters · 7 assumptions · 0 invented entities

The central empirical results depend on several hand-chosen hyperparameters (T, TAD temperature, threshold) and on design axioms (STE, TAD, tie-breaking) that are not derived. No free parameters are fitted to test data; they are training/architecture choices. No new physical entities are postulated.

free parameters (4)
  • Maximum training timestep T = 4 for main results; 2/4/6/8 in scalability study
    Chosen by hand; sets the latency/accuracy trade-off and bounds the temporal resolution of the encoder t_s = ceil((1-x)T).
  • TAD temperature tau = 2
    Chosen by hand; controls the sharpness of confidence weighting in Eq. (13); not swept in main experiments.
  • Membrane time constant / leaky factor tau = not reported
    LIF leak parameter in Eq. (2); influences spike timing and is not specified, so results depend on an unreported value.
  • Firing threshold V_th = not reported (assumed 1)
    Threshold in Eq. (3); standard default but not stated; affects spike timing, energy estimates, and classification decisions.
assumptions (7)
  • domain assumption LIF dynamics discretized as U_l[t] = tau U_l[t-1] + W_l S_{l-1}[t] with soft reset (Eqs. 2-4)
    Adopted as the neuron model; the temporal behavior of the whole framework depends on this discrete update and subtractive reset.
  • ad hoc to paper Latency mapping t_s(x) = ceil((1-x)T) (Eq. 8)
    Defines how feature values become spike times; not derived from optimality or biology.
  • ad hoc to paper Straight-through estimator gradient for the latency encoder (Eq. 9)
    Replaces the non-differentiable encoder with identity in the backward pass; standard practice but an approximation with no error bound.
  • ad hoc to paper Earliest-spike decision with membrane-potential tie-break (Eqs. 10-11)
    Assumes the highest membrane potential at the earliest spike time identifies the true continuous-time first-spike winner.
  • ad hoc to paper TAD confidence weighting (Eqs. 12-14)
    Heuristic loss; no theoretical guarantee that it achieves an optimal speed-accuracy trade-off.
  • ad hoc to paper Unspecified surrogate gradient for the firing function H(.) in Eq. (3)
    The paper says it uses BPTT with surrogate gradients but does not give the surrogate derivative; results depend on this unspecified choice.
  • domain assumption 45nm CMOS energy coefficients (0.9 pJ/AC, 4.6 pJ/MAC) and TrueNorth/SpiNNaker static-dynamic coefficients from [31]
    Energy claims inherit external hardware assumptions and are only rough estimates, as the paper acknowledges.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Latency Coding for Efficient and Low-Latency Deep Spiking Neural Networks." pith.science (2026). https://pith.science/paper/EFOYZZBO

@misc{pith2026260323206,
  author       = {Pith},
  title        = {Pith review of: Latency Coding for Efficient and Low-Latency Deep Spiking Neural Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EFOYZZBO}},
  note         = {Machine review of arXiv:2603.23206}
}
read the original abstract

Spiking neural networks (SNNs) offer a biologically inspired computing paradigm with significant potential for energy-efficient neural processing. Among neural coding schemes of SNNs, Time-To-First-Spike (TTFS) coding, which encodes information through the precise timing of a neuron's first spike, provides exceptional activity sparsity and energy efficiency. However, existing TTFS models lack efficient training methods, suffering from high inference latency and limited performance, limiting their practicality on neuromorphic hardware. In this work, we propose latency coding, an extension of TTFS coding, and present a compatible framework that enables the efficient training of deep latency-coded SNNs by leveraging backpropagation through time (BPTT) algorithm. The framework includes: (1) a latency encoding (LE) module with feature extraction and straight-through estimators to address severe information loss in direct intensity-to-latency mapping; (2) relaxation of the strict single-spike constraint in intermediate layers to improve information propagation and gradient flow; and (3) a temporal adaptive decision (TAD) loss function that dynamically weights supervision signals based on the model's confidence, balancing the trade-off between speed and accuracy. Experimental results demonstrate that our method achieves competitive or superior accuracy compared with existing TTFS-coded SNNs with ultra-low inference latency and high energy efficiency. Latency-coded SNNs also demonstrate improved robustness against input perturbations. These findings highlight latency coding as a practical and hardware-friendly approach for fast and energy-efficient neuromorphic processing.

Figures

Figures reproduced from arXiv: 2603.23206 by the authors.

Figure 1
Figure 1. Overall Framework Subsequently, the latency encoder LE(·) transforms the rescaled features F into spike trains X incorporating a tem￾poral dimension, in which values determine the spike timing: X = LE(F) = [X[1], · · · , X[T]] ∈ {0, 1} T ×C×H×W , (6) Xc,h,w[t] = ( 1, t = ts(Xc,h,w[t]) 0, otherwise , ∀c, h, w , (7) where Xc,h,w denotes the spike train of a single neuron in the feature map, T is the maximum time step.… view at source ↗
Figure 2
Figure 2. Performance evolution across severity levels on CIFAR-100-C. (a) Comparison of mCE from Severity 1 to 5. (b)-(f) Radar charts illustrating the error [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Temporal Similarity Matrix. Calculated on CIFAR100 test set. (a)(b) [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (2 more)
Figure 5
Figure 5. Figure 5: Distribution of first spike time against sample difficulty. (a) Training [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: Evaluation of time scalability robustness. The plots show the classification accuracy (blue lines) and actual average inference timesteps on test sets [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

56 extracted references · 10 linked inside Pith

  1. [34]

    High-performance deep spiking neural networks with 0.3 spikes per neuron,

    A. Stanojevic, S. Wo ´zniak, G. Bellec, G. Cherubini, A. Pantazi, and W. Gerstner, “High-performance deep spiking neural networks with 0.3 spikes per neuron,”Nature Communications, vol. 15, no. 1, p. 6793, 2024

  2. [1]

    Optimal ann-snn conversion for fast and accurate inference in deep spiking neural networks,

    J. Ding, Z. Yu, Y . Tian, and T. Huang, “Optimal ann-snn conversion for fast and accurate inference in deep spiking neural networks,”arXiv preprint arXiv:2105.11654, 2021

  3. [2]

    Attention-based deep spiking neural networks for temporal credit assignment problems,

    L. Qin, Z. Wang, R. Yan, and H. Tang, “Attention-based deep spiking neural networks for temporal credit assignment problems,”IEEE Trans- actions on Neural Networks and Learning Systems, vol. 35, no. 8, pp. 10 301–10 311, 2023

  4. [3]

    A tandem learning rule for effective training and rapid inference of deep spiking neural networks,

    J. Wu, Y . Chua, M. Zhang, G. Li, H. Li, and K. C. Tan, “A tandem learning rule for effective training and rapid inference of deep spiking neural networks,”IEEE Transactions on Neural Networks and Learning Systems, vol. 34, no. 1, pp. 446–460, 2021

  5. [4]

    End to end learning of spiking neural network based on r- stdp for a lane keeping vehicle,

    Z. Bing, C. Meschede, K. Huang, G. Chen, F. Rohrbein, M. Akl, and A. Knoll, “End to end learning of spiking neural network based on r- stdp for a lane keeping vehicle,” in2018 IEEE international conference on robotics and automation (ICRA). IEEE, 2018, pp. 4725–4732

  6. [5]

    Indirect and direct training of spiking neural networks for end-to-end control of a lane-keeping vehicle,

    Z. Bing, C. Meschede, G. Chen, A. Knoll, and K. Huang, “Indirect and direct training of spiking neural networks for end-to-end control of a lane-keeping vehicle,”Neural Networks, vol. 121, pp. 21–36, 2020

  7. [6]

    Progressive tandem learning for pattern recognition with deep spiking neural networks,

    J. Wu, C. Xu, X. Han, D. Zhou, M. Zhang, H. Li, and K. C. Tan, “Progressive tandem learning for pattern recognition with deep spiking neural networks,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 11, pp. 7824–7840, 2021

  8. [7]

    A quantitative description of mem- brane current and its application to conduction and excitation in nerve

    A. L. Hodgkin and A. F. Huxley, “A quantitative description of mem- brane current and its application to conduction and excitation in nerve.” Journal of Physiology, vol. 117, 1952

Show all 56 references
  1. [8]

    Networks of spiking neurons: the third generation of neural network models,

    W. Maass, “Networks of spiking neurons: the third generation of neural network models,”Neural Networks, vol. 10, no. 9, pp. 1659–1671, 1997

  2. [9]

    Direct training for spiking neural networks: Faster, larger, better,

    Y . Wu, L. Deng, G. Li, J. Zhu, Y . Xie, and L. Shi, “Direct training for spiking neural networks: Faster, larger, better,” inProceedings of the AAAI conference on artificial intelligence, vol. 33, no. 01, 2019, pp. 1311–1318

  3. [10]

    Mapping from frame- driven to frame-free event-driven vision systems by low-rate rate coding and coincidence processing–application to feedforward convnets,

    J. A. P ´erez-Carrasco, B. Zhao, C. Serrano, B. Acha, T. Serrano- Gotarredona, S. Chen, and B. Linares-Barranco, “Mapping from frame- driven to frame-free event-driven vision systems by low-rate rate coding and coincidence processing–application to feedforward convnets,”IEEE t...

  4. [11]

    Rmp-snn: Residual membrane potential neuron for enabling deeper high-accuracy and low-latency spiking neural network,

    B. Han, G. Srinivasan, and K. Roy, “Rmp-snn: Residual membrane potential neuron for enabling deeper high-accuracy and low-latency spiking neural network,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 13 558–13 567

  5. [12]

    Optimal ann- snn conversion for high-accuracy and ultra-low-latency spiking neural networks,

    T. Bu, W. Fang, J. Ding, P. Dai, Z. Yu, and T. Huang, “Optimal ann- snn conversion for high-accuracy and ultra-low-latency spiking neural networks,”arXiv preprint arXiv:2303.04347, 2023

  6. [13]

    Surrogate gradient learning in spiking neural networks: Bringing the power of gradient-based opti- mization to spiking neural networks,

    E. O. Neftci, H. Mostafa, and F. Zenke, “Surrogate gradient learning in spiking neural networks: Bringing the power of gradient-based opti- mization to spiking neural networks,”IEEE Signal Processing Magazine, vol. 36, no. 6, pp. 51–63, 2019

  7. [14]

    Spatio-temporal backpropa- gation for training high-performance spiking neural networks,

    Y . Wu, L. Deng, G. Li, J. Zhu, and L. Shi, “Spatio-temporal backpropa- gation for training high-performance spiking neural networks,”Frontiers in neuroscience, vol. 12, p. 331, 2018

  8. [15]

    Assisting training of deep spiking neural networks with parameter initialization,

    J. Ding, J. Zhang, T. Huang, J. K. Liu, and Z. Yu, “Assisting training of deep spiking neural networks with parameter initialization,”IEEE Transactions on Neural Networks and Learning Systems, 2025

  9. [16]

    Spike arrival times: A highly efficient coding scheme for neural networks,

    S. J. Thorpe, “Spike arrival times: A highly efficient coding scheme for neural networks,”Parallel processing in neural systems, pp. 91–94, 1990

  10. [17]

    Temporal encoding of two- dimensional patterns by single units in primate inferior temporal cortex. iii. information theoretic analysis,

    L. M. Optican and B. J. Richmond, “Temporal encoding of two- dimensional patterns by single units in primate inferior temporal cortex. iii. information theoretic analysis,”Journal of neurophysiology, vol. 57, no. 1, pp. 162–178, 1987

  11. [18]

    Information encoding and the responses of single neurons in the primate temporal visual cortex,

    M. J. Tovee, E. T. Rolls, A. Treves, and R. P. Bellis, “Information encoding and the responses of single neurons in the primate temporal visual cortex,”Journal of neurophysiology, vol. 70, no. 2, pp. 640–654, 1993

  12. [19]

    Decoding cortical neuronal signals: network models, information estimation and spatial tuning,

    T. W. Kjaer, J. A. Hertz, and B. J. Richmond, “Decoding cortical neuronal signals: network models, information estimation and spatial tuning,”Journal of computational neuroscience, vol. 1, no. 1, pp. 109– 139, 1994

  13. [20]

    Information encoding in short firing rate epochs by single neurons in the primate temporal visual cortex,

    M. J. Tovee and E. T. Rolls, “Information encoding in short firing rate epochs by single neurons in the primate temporal visual cortex,”Visual cognition, vol. 2, no. 1, pp. 35–58, 1995

  14. [21]

    Stdp-based spiking deep convolutional neural networks for object recognition,

    S. R. Kheradpisheh, M. Ganjtabesh, S. J. Thorpe, and T. Masquelier, “Stdp-based spiking deep convolutional neural networks for object recognition,”Neural Networks, vol. 99, pp. 56–67, 2018

  15. [22]

    Precise-spike-driven synaptic plasticity: Learning hetero-association of spatiotemporal spike patterns,

    Q. Yu, H. Tang, K. C. Tan, and H. Li, “Precise-spike-driven synaptic plasticity: Learning hetero-association of spatiotemporal spike patterns,” Plos one, vol. 8, no. 11, p. e78318, 2013

  16. [23]

    Supervised learning based on temporal coding in spiking neural networks,

    H. Mostafa, “Supervised learning based on temporal coding in spiking neural networks,”IEEE transactions on neural networks and learning systems, vol. 29, no. 7, pp. 3227–3235, 2017

  17. [24]

    Rectified linear postsynaptic potential function for backpropagation in deep spiking neural networks,

    M. Zhang, J. Wang, J. Wu, A. Belatreche, B. Amornpaisannon, Z. Zhang, V . P. K. Miriyala, H. Qu, Y . Chua, T. E. Carlsonet al., “Rectified linear postsynaptic potential function for backpropagation in deep spiking neural networks,”IEEE transactions on neural networks and learn...

  18. [25]

    Temporal-coded spiking neural networks with dynamic firing threshold: Learning with event-driven backpropagation,

    W. Wei, M. Zhang, H. Qu, A. Belatreche, J. Zhang, and H. Chen, “Temporal-coded spiking neural networks with dynamic firing threshold: Learning with event-driven backpropagation,” inProceedings of the IEEE/CVF international conference on computer vision, 2023, pp. 10 552–10 562

  19. [26]

    A survey of encoding techniques for signal processing in spiking neural networks,

    D. Auge, J. Hille, E. Mueller, and A. Knoll, “A survey of encoding techniques for signal processing in spiking neural networks,”Neural Processing Letters, no. 5, 2021

  20. [27]

    Neural coding in spiking neural networks: A comparative study for robust neuromorphic systems

    W. Guo, M. E. Fouda, A. Eltawil, and K. N. Salama, “Neural coding in spiking neural networks: A comparative study for robust neuromorphic systems.”Frontiers in Neuroscience, 2021

  21. [28]

    Batch normalization: Accelerating deep network training by reducing internal covariate shift,

    S. Ioffe and C. Szegedy, “Batch normalization: Accelerating deep network training by reducing internal covariate shift,”JMLR.org, 2015

  22. [29]

    Adaptive computation time for recurrent neural networks,

    A. Graves, “Adaptive computation time for recurrent neural networks,” arXiv preprint arXiv:1603.08983, 2016

  23. [30]

    Pondernet: Learning to ponder,

    A. Banino, J. Balaguer, and C. Blundell, “Pondernet: Learning to ponder,”arXiv preprint arXiv:2107.05407, 2021

  24. [31]

    T2fsnn: deep spiking neural networks with time-to-first-spike coding,

    S. Park, S. Kim, B. Na, and S. Yoon, “T2fsnn: deep spiking neural networks with time-to-first-spike coding,” in2020 57th ACM/IEEE design automation conference (DAC). IEEE, 2020, pp. 1–6

  25. [32]

    Deep spiking neural network: Energy efficiency through time based coding,

    B. Han and K. Roy, “Deep spiking neural network: Energy efficiency through time based coding,” inEuropean conference on computer vision. Springer, 2020, pp. 388–404

  26. [33]

    Training energy-efficient deep spiking neural net- works with time-to-first-spike coding,

    S. Park and S. Yoon, “Training energy-efficient deep spiking neural net- works with time-to-first-spike coding,”arXiv preprint arXiv:2106.02568, 2021

  27. [35]

    Temporal-coded spiking Transformer,

    Q. Sun, C. Lu, W. Chen, W. Wei, J. Wang, J. Zhang, X. Liu, Y . Ye, Y . Yang, and M. Zhang, “Temporal-coded spiking Transformer,” inProc. 33rd ACM Int. Conf. Multimedia (MM), 2025, pp. 2616–2624

  28. [36]

    Very deep convolutional networks for large-scale image recognition,

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

  29. [37]

    Deep residual learning in spiking neural networks,

    W. Fang, Z. Yu, Y . Chen, T. Huang, T. Masquelier, and Y . Tian, “Deep residual learning in spiking neural networks,”Advances in Neural Information Processing Systems, vol. 34, pp. 21 056–21 069, 2021

  30. [38]

    Decoupled weight decay regularization,

    I. Loshchilov and F. Hutter, “Decoupled weight decay regularization,” arXiv preprint arXiv:1711.05101, 2017

  31. [39]

    Autoaug- ment: Learning augmentation strategies from data,

    E. D. Cubuk, B. Zoph, D. Mane, V . Vasudevan, and Q. V . Le, “Autoaug- ment: Learning augmentation strategies from data,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 113–123

  32. [40]

    Tiny imagenet visual recognition challenge,

    Y . Le and X. Yang, “Tiny imagenet visual recognition challenge,”CS 231N, vol. 7, no. 7, p. 3, 2015

  33. [41]

    Randaugment: Practical automated data augmentation with a reduced search space,

    E. D. Cubuk, B. Zoph, J. Shlens, and Q. V . Le, “Randaugment: Practical automated data augmentation with a reduced search space,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition workshops, 2020, pp. 702–703

  34. [42]

    Cifar10-dvs: an event-stream dataset for object classification,

    H. Li, H. Liu, X. Ji, G. Li, and L. Shi, “Cifar10-dvs: an event-stream dataset for object classification,”Frontiers in neuroscience, vol. 11, p. 244131, 2017

  35. [43]

    Neuromorphic data augmentation for training spiking neural networks,

    Y . Li, Y . Kim, H. Park, T. Geller, and P. Panda, “Neuromorphic data augmentation for training spiking neural networks,” inEuropean Conference on Computer Vision. Springer, 2022, pp. 631–649

  36. [44]

    Diet-snn: A low-latency spiking neural network with direct input encoding and leakage and threshold optimization,

    N. Rathi and K. Roy, “Diet-snn: A low-latency spiking neural network with direct input encoding and leakage and threshold optimization,” IEEE Transactions on Neural Networks and Learning Systems, vol. 34, no. 6, pp. 3174–3182, 2021

  37. [45]

    Spike-based dynamic computing with asynchronous sensing-computing neuromorphic chip,

    M. Yao, O. Richter, G. Zhao, N. Qiao, Y . Xing, D. Wang, T. Hu, W. Fang, T. Demirci, M. De Marchiet al., “Spike-based dynamic computing with asynchronous sensing-computing neuromorphic chip,” Nature Communications, vol. 15, no. 1, p. 4464, 2024

  38. [46]

    Pruning convolutional neural networks for resource efficient inference,

    P. Molchanov, S. Tyree, T. Karras, T. Aila, and J. Kautz, “Pruning convolutional neural networks for resource efficient inference,”arXiv preprint arXiv:1611.06440, 2016

  39. [47]

    A million spiking-neuron integrated circuit with a scalable communication network and interface,

    P. A. Merolla, J. V . Arthur, R. Alvarez-Icaza, A. S. Cassidy, J. Sawada, F. Akopyan, B. L. Jackson, N. Imam, C. Guo, Y . Nakamuraet al., “A million spiking-neuron integrated circuit with a scalable communication network and interface,”Science, vol. 345, no. 6197, pp. 668–673, 2014

  40. [48]

    A hybrid neural coding approach for pattern recognition with spiking neural networks,

    X. Chen, Q. Yang, J. Wu, H. Li, and K. C. Tan, “A hybrid neural coding approach for pattern recognition with spiking neural networks,”IEEE transactions on pattern analysis and machine intelligence, vol. 46, no. 5, pp. 3064–3078, 2023

  41. [49]

    Accurate and efficient time-domain classification with adaptive spiking recurrent neural networks,

    B. Yin, F. Corradi, and S. M. Boht ´e, “Accurate and efficient time-domain classification with adaptive spiking recurrent neural networks,”Nature Machine Intelligence, vol. 3, no. 10, pp. 905–913, 2021

  42. [50]

    Toward scalable, efficient, and accurate deep spiking neural networks with backward residual connec- tions, stochastic softmax, and hybridization,

    P. Panda, S. A. Aketi, and K. Roy, “Toward scalable, efficient, and accurate deep spiking neural networks with backward residual connec- tions, stochastic softmax, and hybridization,”Frontiers in Neuroscience, vol. 14, p. 653, 2020

  43. [51]

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

    M. Horowitz, “1.1 computing’s energy problem (and what we can do about it),” in2014 IEEE international solid-state circuits conference digest of technical papers (ISSCC). IEEE, 2014, pp. 10–14

  44. [52]

    Temporal efficient training of spiking neural network via gradient re-weighting,

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

  45. [53]

    The spinnaker project,

    S. B. Furber, F. Galluppi, S. Temple, and L. A. Plana, “The spinnaker project,”Proceedings of the IEEE, vol. 102, no. 5, pp. 652–665, 2014

  46. [54]

    Benchmarking neural network ro- bustness to common corruptions and perturbations,

    D. Hendrycks and T. Dietterich, “Benchmarking neural network ro- bustness to common corruptions and perturbations,”arXiv preprint arXiv:1903.12261, 2019

  47. [55]

    Unleashing the potential of spik- ing neural networks with dynamic confidence,

    C. Li, E. G. Jones, and S. Furber, “Unleashing the potential of spik- ing neural networks with dynamic confidence,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 13 350–13 360

  48. [56]

    Seenn: Towards temporal spiking early exit neural networks,

    Y . Li, T. Geller, Y . Kim, and P. Panda, “Seenn: Towards temporal spiking early exit neural networks,”Advances in Neural Information Processing Systems, vol. 36, pp. 63 327–63 342, 2023

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.