Pith. sign in

REVIEW 4 major objections 4 minor 57 references

Efficient Speech Command Recognition Leveraging Spiking Neural Network and Curriculum Learning-based Knowledge Distillation

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

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

Pith's one-line read This paper claims a fully spike-driven speech recognizer can match state-of-the-art accuracy while running on 60% fewer time steps through curriculum-based knowledge distillation.

desk verdict Solid SNN accuracy engineering with a genuinely useful curriculum distillation idea, but the headline energy number is an operation-count claim that doesn't yet hold up. read the letter →

arxiv 2412.12858 v1 pith:SH6DCGNI submitted 2024-12-17 cs.LG cs.AIcs.HC

Jiaqi Wang , Liutao Yu , Liwei Huang , Chenlin Zhou , Han Zhang , Zhenxi Song , Min Zhang , Zhengyu Ma
show 1 more author
Zhiguo Zhang
This is my paper · ORCID
classification cs.LGcs.AIcs.HC
keywords spikingneuralnetworksspeechcommandrecognitionknowledgedistillationcurriculumlearningself-attentionrotarypositionembeddinggatedconvolutionenergy-efficientedgecomputing
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 claims that a fully spike-driven speech-command recognizer can match or beat state-of-the-art spiking networks while running on far fewer time steps. The model, SpikeSCR, combines a spiking self-attention branch with rotary position encoding for global context and a separable gated convolution branch for local context, so accuracy keeps climbing as time steps grow instead of saturating. To make short-time-step operation cheap, the authors add KDCL, a curriculum knowledge-distillation schedule that first trains a long-time-step teacher and then progressively distills to students at 200, 100, and finally 40 time steps. At 40 time steps the distilled model reports 93.60% on SHD, 80.25% on SSC, and 95.01% on GSC, matching or beating prior methods that use 100 steps, with 60% fewer time steps and 54.8% lower estimated energy. The result matters because it points to a way to deploy temporal spiking models on energy-limited edge devices without sacrificing accuracy.

What carries the argument

The load-bearing mechanism is the pairing of a spike-domain rotary position encoding with a curriculum of distillation teachers. Instead of injecting float-valued sinusoidal position vectors, SpikeSCR applies rotary position embeddings through a Hadamard product followed immediately by a LIF spike neuron, so positional information enters the spiking self-attention without converting the whole feature map back to real-valued multiply-accumulate computation. The local branch uses a spiking gated unit that gates one split of the features with the spiking output of the other split, keeping operations as spike-based accumulates. KDCL is the second half of the machinery: a teacher trained on long time steps is frozen and its logits distilled into a student at shorter time steps, then the student becomes the next teacher, so knowledge moves down the curriculum ladder one rung at a time.

What would settle it

Take the two-block SpikeSCR at 40 and 100 time steps on the SSC test set, run it on a neuromorphic chip or a cycle-accurate power simulator, and compare measured per-command energy to the reported 0.0142 mJ and 0.0314 mJ; a reduction far from 54.8%, or a cost breakdown showing the rotary-position MACs dominate, would refute the energy claim.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that the accuracy-latency trade-off of a spiking network in speech command recognition can be moved by architecture and by curriculum distillation rather than by adding parameters. SpikeSCR is a global-local hybrid: the global path is spiking self-attention with rotary position embeddings applied inside the spike domain, and the local path is a separable convolution with a spiking gated unit that keeps computation in sparse spike-accumulate operations. Trained directly, it outperforms prior spiking networks on SHD, SSC, and GSC at equal time steps, and its accuracy keeps improving as time steps grow to 400-500 instead of saturating. KDCL then turns the long-time-step model into a teacher, distilling it stepwise to 200, 100, and finally 40 time steps; the paper reports 60% fewer time steps and a 54.8% drop in estimated energy while accuracy stays within a point or two of 100-step state-of-the-art results.

Load-bearing premise

The energy-reduction claim rests on a cost model that counts only spike-based accumulate operations at a fixed 0.9 pJ each, while ignoring memory access, control overhead, and the real-valued rotary position embedding that is applied before the next spiking layer.

Editorial extensions

If this is right

  • At 40 time steps, the distilled SpikeSCR is competitive with 100-step state-of-the-art models on all three benchmarks, so always-on keyword spotting can run with lower latency and a smaller energy budget.
  • Because SpikeSCR accuracy still rises at 400-500 time steps, it can serve as a stable teacher for distillation, a role that the dilated-convolution baseline cannot fill since its accuracy declines at long time steps.
  • The combination of rotary position embedding followed by a LIF neuron shows that positional information can be added in the spike domain without switching to float multiply-accumulate operations.
  • The KDCL schedule is not tied to speech: any temporal SNN trained with long time steps can be chained through successive shorter-time-step students to compress latency with modest accuracy loss.

Reading between the lines

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

  • A natural next experiment, not run in the paper, would apply KDCL to event-based audio or video with more than two curricula to see whether the 60% time-step reduction generalizes or saturates.
  • Recomputing the energy ledger with the multiply-accumulate operations hidden in the rotary position embedding and with memory-access costs would likely raise the absolute energy numbers; the qualitative ranking of 40-step versus 100-step models may survive, but the 54.8% figure probably would not.
  • The ablation pattern suggests that on Mel-spectrogram inputs the local separable gated convolution carries more of the accuracy than the global attention branch, a ranking worth testing on other audio features such as raw waveforms.
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

4 major / 4 minor

Summary. The paper proposes SpikeSCR, a fully spike-driven SNN architecture for speech command recognition that combines spiking self-attention with rotary position embedding (global branch) and a separable gated convolution module (local branch), together with a curriculum-learning-based knowledge distillation method (KDCL) that transfers knowledge from models trained with long time steps to students trained with shorter time steps. The authors evaluate on SHD, SSC, and GSC and report that SpikeSCR outperforms prior SNN methods at matched time steps, and that KDCL allows a 60% reduction in time steps and a 54.8% reduction in energy while keeping performance comparable to recent SOTA.

Significance. If the main empirical claims hold, the paper makes a useful contribution: the architecture is a reasonable synthesis of spiking attention, rotary position encoding, and gated convolutions, and the KDCL schedule is a practical way to recover accuracy at short time steps. The systematic ablations (Tables 3, 9, 10), the matched-time-step comparisons with DCLS (Tables 1, 7, 8), and the use of the SpikingJelly framework are strengths. However, the headline energy-reduction number rests on an idealized operation-count model that omits real-valued operations and memory costs, and several key accuracy comparisons are reported without variance estimates. These issues are load-bearing for the abstract's central claims and need to be addressed before the paper can be accepted.

major comments (4)
  1. [Appendix D, Tables 7 and 8] The energy model in Eqs. (19)-(20) counts only MAC operations for the first Conv layer and AC operations for all subsequent layers, but the RoPE operation in Eq. (6), Q'_S = SN_RQ(R_Q ⊙ Q_S), involves an elementwise Hadamard product between a binary spike tensor Q_S and a real-valued rotary matrix R_Q. This operation is not a spike-based AC operation, and it produces real-valued intermediate activations before the LIF neuron re-spikifies them. The cost of this operation, including memory traffic, is absent from the energy accounting, yet it is precisely the operation introduced to improve accuracy. Because the abstract's 54.8% energy reduction is computed from these tables, the headline number is at best an operation-count reduction, not a hardware energy reduction. The authors should either extend the model to include RoPE's real-valued multiplies and memory costs, or explicitly present the numbers as theoretical AC/MAC counts with a clear caveat that real hardware energy will be higher.
  2. [Algorithm 1] The distillation step is not well-defined as written. In Algorithm 1, the teacher receives mini-batch x1(i) from D1 and the student receives mini-batch x2(i) from D2, and then LKD is computed between OT1(i) and OS1(i). For KL-divergence distillation to be meaningful, x1(i) and x2(i) must be paired versions of the same utterance under the easy and hard curricula. The text only says "D1 and D2 synchronization" without explaining how the two data streams are aligned. If they are not paired, the distillation objective compares teacher predictions on one input with student predictions on a different input, which does not transfer per-sample knowledge and may actually harm training. The authors must state explicitly how synchronization is performed and confirm that both models see the same underlying samples at each iteration.
  3. [Tables 1, 3, 7, 8] All accuracy results are reported as single numbers without error bars or multiple seeds. Several of the most important comparisons are very close, for example SSC 80.25% vs. DCLS 80.16% and GSC 95.01% vs. DCLS 95.00% in Table 1. With a single run it is not possible to know whether these differences are statistically meaningful, yet the abstract and main text describe these as clear SOTA improvements. At minimum, the authors should report mean and standard deviation over at least three independent training runs for the main configurations, and temper the SOTA language if the margins are within noise.
  4. [Tables 7 and 8] The reported energy consumption of the DCLS baseline decreases as the number of time steps increases (Table 7: 0.0276 mJ at 40 steps, 0.0216 at 100, 0.0118 at 200, 0.0110 at 500; Table 8 shows a similar but milder trend). Since longer time steps normally imply more temporal processing, this trend is surprising and suggests that the SOP counting, firing-rate estimation, or input preprocessing is doing something non-obvious for DCLS. The paper's explanation in Appendix C that DCLS "mitigates rising energy consumption by learning delays with only a few kernel counts" is not quantitatively justified. The authors should either explain the mechanism with measured firing rates or acknowledge that the comparison is not apples-to-apples in terms of how time steps affect each architecture.
minor comments (4)
  1. [Table 1] The caption says "† indicates the performance after KDCL" but the entries use a slash notation such as "93.60 †/92.01"; the meaning of the number before and after the slash is not explained and should be clarified.
  2. [Appendix E] The code availability statement says the organized code "will be made publicly available upon reaching the camera-ready version." For a paper whose main claims are empirical and whose comparisons depend on exact preprocessing and energy accounting, releasing code and random seeds at submission time, or at least providing a detailed configuration file, would substantially improve reproducibility.
  3. [Section KDCL] The loss weight λ2 = 0.5 is described as "refined through experiments" but no sensitivity analysis is shown. A short ablation over λ2 and the distillation temperature τ would help the reader judge how robust the KDCL gains are to hyperparameter choice.
  4. [Appendix D] Equation (18) defines SOP^l = fr^{l-1} × FLOP^l but does not state whether the firing rate is averaged over time steps or per simulation, nor whether FLOP^l is per time step or over the full sequence. This ambiguity makes it difficult to reproduce the energy numbers in Tables 7 and 8.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the results are empirical and the energy estimates follow a stated external cost model.

full rationale

The paper's central claims are empirical: SpikeSCR accuracy numbers are measured on three benchmarks, and the KDCL results compare trained student models against trained teachers. No prediction is derived by construction from a fitted parameter: the KDCL loss weights (lambda1 = 1.0, lambda2 = 0.5) are validation-tuned hyperparameters, not inputs that force the reported accuracies. The 54.8% energy reduction is computed from a transparent operation-count model (Appendix D, Eqs. 18-20) using fixed hardware cost constants (E_MAC = 4.6 pJ, E_AC = 0.9 pJ at 45nm). That the reduction follows from fewer time steps and fewer spike operations is the intended meaning of the claim, not a hidden equivalence between input and output. The paper cites prior work by overlapping authors (e.g., Spikformer for SSA, SGLFormer as one of several references for the energy-calculation convention), but these citations are not load-bearing axioms; the architecture component SSA is used as a building block and the energy formula is standard in the SNN literature and independently grounded in Horowitz's 45nm data. The skeptic's concern that the energy model omits memory access, control overhead, and real-valued RoPE operations is a correctness or realism concern about the energy metric, not circularity: under the paper's own stated model, the energy numbers are legitimate, and the paper does not disguise a fitted quantity as a prediction. Therefore, no circular step meeting the required evidentiary standard is present.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The central empirical claim rests on standard SNN training assumptions, a theoretical energy model, and a small set of tuned hyperparameters. None of the components require new physical entities; the main unstated cost is in the energy accounting for rotary embeddings and memory overhead.

free parameters (6)
  • KDCL loss weight lambda2 = 0.5
    Set to 0.5 after experiments, as stated in the Training Strategy section; balances cross-entropy and distillation terms and directly affects KDCL accuracy.
  • Curriculum time-step schedule = SHD: 100, 40; SSC/GSC: 500, 200, 100, 40
    Chosen per dataset; the 60% time-step reduction claim (100 to 40) is defined by this schedule, which is tuned rather than derived.
  • EventDrop augmentation parameters = drop proportion 0.5, time drop 0.2 for SHD and 0.1 for SSC, neuron drop 20 for SHD and 10 for SSC
    Manual augmentation settings given in Appendix A; these affect the trained model accuracy.
  • SpecAugment mask parameters = 1 frequency mask of size 10 and 1 time mask of size 0.25
    Manual augmentation settings for GSC, given in Appendix A.
  • LIF neuron parameters = tau = 2.0, V_threshold = 1.0
    Fixed hyperparameters used for all experiments, listed in Table 6; part of the spiking neuron model.
  • Architecture dimensions = 1 or 2 SGLE blocks, 8 or 16 heads, hidden sizes 128 or 256, depthwise kernel 31
    Model capacity choices compared in Table 1; the SOTA claims depend on these sizes, which are selected by hand.
assumptions (4)
  • domain assumption LIF neuron dynamics and surrogate-gradient BPTT can train the network end-to-end.
    Equations (1) to (3) and the Methods section; standard in SNN literature, not re-derived here.
  • domain assumption Energy cost is fully captured by SOP counts with E_AC = 0.9 pJ and E_MAC = 4.6 pJ on 45nm hardware, ignoring memory and control overhead.
    Appendix D, Eqs. (18) to (20); the central energy-reduction claim rests on this accounting.
  • ad hoc to paper The rotary position embedding applied to spike tensors can be implemented without additional MAC operations beyond the counted AC operations.
    Eq. (6) applies a real-valued rotary matrix R to spike vectors Q_S with a Hadamard product, then spikifies; the energy model in Appendix D does not add a term for this float-valued operation.
  • domain assumption Using the DCLS preprocessing pipeline, including spatio-temporal binning of 700 input neurons into 140, preserves comparability with prior SOTA results.
    Section Experiments; the accuracy comparisons inherit this assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Efficient Speech Command Recognition Leveraging Spiking Neural Network and Curriculum Learning-based Knowledge Distillation." pith.science (2026). https://pith.science/paper/SH6DCGNI

@misc{pith2026241212858,
  author       = {Pith},
  title        = {Pith review of: Efficient Speech Command Recognition Leveraging Spiking Neural Network and Curriculum Learning-based Knowledge Distillation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SH6DCGNI}},
  note         = {Machine review of arXiv:2412.12858}
}
read the original abstract

The intrinsic dynamics and event-driven nature of spiking neural networks (SNNs) make them excel in processing temporal information by naturally utilizing embedded time sequences as time steps. Recent studies adopting this approach have demonstrated SNNs' effectiveness in speech command recognition, achieving high performance by employing large time steps for long time sequences. However, the large time steps lead to increased deployment burdens for edge computing applications. Thus, it is important to balance high performance and low energy consumption when detecting temporal patterns in edge devices. Our solution comprises two key components. 1). We propose a high-performance fully spike-driven framework termed SpikeSCR, characterized by a global-local hybrid structure for efficient representation learning, which exhibits long-term learning capabilities with extended time steps. 2). To further fully embrace low energy consumption, we propose an effective knowledge distillation method based on curriculum learning (KDCL), where valuable representations learned from the easy curriculum are progressively transferred to the hard curriculum with minor loss, striking a trade-off between power efficiency and high performance. We evaluate our method on three benchmark datasets: the Spiking Heidelberg Dataset (SHD), the Spiking Speech Commands (SSC), and the Google Speech Commands (GSC) V2. Our experimental results demonstrate that SpikeSCR outperforms current state-of-the-art (SOTA) methods across these three datasets with the same time steps. Furthermore, by executing KDCL, we reduce the number of time steps by 60% and decrease energy consumption by 54.8% while maintaining comparable performance to recent SOTA results. Therefore, this work offers valuable insights for tackling temporal processing challenges with long time sequences in edge neuromorphic computing systems.

Figures

Figures reproduced from arXiv: 2412.12858 by the authors.

Figure 1
Figure 1. Method Overview: (a) The pipeline of SpikeSCR. (b) The framework of SpikeSCR consists of a spike/spec aug [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 3
Figure 3. Illustration of our proposed KDCL. Taking the [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Comparative analysis of model performance and energy consumption from short to long time steps on SSC dataset. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figures from the paper (7 more)
Figure 5
Figure 5. Figure 5: Comparing the performance of direct training, [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Firing rate of the LIF Neuron SN2 in SGU on SHD, SSC and GSC datasets. removed from the model, leading to performance reductions. The removal of the KDCL method results in a 0.34% de￾crease in accuracy. Further, eliminating the augment mod￾ule reduces accuracy by an ad…
Figure 7
Figure 7. Figure 7: Comparison of the ”Right” command in GSC [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 8
Figure 8. Figure 8: Comparison of the ”Cat” command in SSC dataset [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 10
Figure 10. Figure 10: Comparative analysis of performance and energy consumption from short to long time steps on GSC dataset. [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]
Figure 11
Figure 11. Figure 11: Comparing the performance of SpikeSCR with [PITH_FULL_IMAGE:figures/full_fig_p012_11.png]
Figure 12
Figure 12. Figure 12: Comparing the performance of SpikeSCR with [PITH_FULL_IMAGE:figures/full_fig_p012_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

57 extracted references · 42 canonical work pages

  1. [1]

    Bittar, A.; and Garner, P. N. 2022. A surrogate gradient spiking baseline for speech command recognition. Frontiers in Neuroscience, 16: 865897

    work page 2022

  2. [2]

    Bu, T.; Fang, W.; Ding, J.; DAI, P.; Yu, Z.; and Huang, T. 2022. Optimal ANN - SNN Conversion for High-accuracy and Ultra-low-latency Spiking Neural Networks. In International Conference on Learning Representations

    2022

  3. [3]

    Chen, G.; Peng, P.; Li, G.; and Tian, Y. 2023. Training full spike neural networks via auxiliary accumulation pathway. arXiv preprint arXiv:2301.11929

    arXiv 2023

  4. [4]

    Cramer, B.; Stradmann, Y.; Schemmel, J.; and Zenke, F. 2020. The heidelberg spiking data sets for the systematic evaluation of spiking neural networks. IEEE Transactions on Neural Networks and Learning Systems, 33(7): 2744--2757

    work page 2020

  5. [5]

    Dampfhoffer, M.; Mesquida, T.; Valentian, A.; and Anghel, L. 2022. Investigating current-based and gating approaches for accurate and energy-efficient spiking recurrent neural networks. In International Conference on Artificial Neural Networks, 359--370. Springer

    work page 2022

  6. [6]

    R.; and Beerel, P

    Datta, G.; Kundu, S.; Jaiswal, A. R.; and Beerel, P. A. 2022. ACE-SNN: Algorithm-hardware co-design of energy-efficient & low-latency deep spiking neural networks for 3d image recognition. Frontiers in Neuroscience, 16: 815258

    work page 2022

  7. [7]

    N.; Fan, A.; Auli, M.; and Grangier, D

    Dauphin, Y. N.; Fan, A.; Auli, M.; and Grangier, D. 2017. Language modeling with gated convolutional networks. In International Conference on Machine Learning, 933--941. PMLR

    work page 2017

  8. [8]

    J.; and Latr \'e , S

    Deckers, L.; Van Damme, L.; Van Leekwijck, W.; Tsang, I. J.; and Latr \'e , S. 2024. Co-learning synaptic delays, weights and adaptation in spiking neural networks. Frontiers in Neuroscience, 18: 1360300

    work page 2024

Show all 57 references
  1. [9]

    Deng, S.; and Gu, S. 2021. Optimal Conversion of Conventional Artificial Neural Networks to Spiking Neural Networks. In International Conference on Learning Representations

    2021

  2. [10]

    Fang, W.; Chen, Y.; Ding, J.; Yu, Z.; Masquelier, T.; Chen, D.; Huang, L.; Zhou, H.; Li, G.; and Tian, Y. 2023. Spikingjelly: An open-source machine learning infrastructure platform for spike-based intelligence. Science Advances, 9(40): eadi1480

    2023

  3. [11]

    Fang, W.; Yu, Z.; Chen, Y.; Huang, T.; Masquelier, T.; and Tian, Y. 2021. Deep Residual Learning in Spiking Neural Networks. In Advances in Neural Information Processing Systems, volume 34, 21056--21069

    2021

  4. [12]

    Gu, F.; Sng, W.; Hu, X.; and Yu, F. 2021. EventDrop: Data Augmentation for Event-based Learning. In 30th International Joint Conference on Artificial Intelligence, IJCAI 202

    2021

  5. [13]

    Gulati, A.; Qin, J.; Chiu, C.-C.; Parmar, N.; Zhang, Y.; Yu, J.; Han, W.; Wang, S.; Zhang, Z.; Wu, Y.; et al. 2020. Conformer: Convolution-augmented transformer for speech recognition. arXiv preprint arXiv:2005.08100

    2020 arXiv

  6. [14]

    Guo, Y.; Chen, Y.; Liu, X.; Peng, W.; Zhang, Y.; Huang, X.; and Ma, Z. 2024. Ternary spike: Learning ternary spikes for spiking neural networks. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, 12244--12252

    2024

  7. [15]

    Hammouamri, I.; Khalfaoui-Hassani, I.; and Masquelier, T. 2024. Learning Delays in Spiking Neural Networks using Dilated Convolutions with Learnable Spacings. In International Conference on Learning Representations

    2024

  8. [16]

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

    2014

  9. [17]

    Kim, K.; Ji, B.; Yoon, D.; and Hwang, S. 2021. Self-knowledge distillation with progressive refinement of targets. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 6567--6576

    2021

  10. [18]

    Kim, Y.; Li, Y.; Park, H.; Venkatesha, Y.; Hambitzer, A.; and Panda, P. 2023. Exploring temporal information dynamics in spiking neural networks. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, 8308--8316

    2023

  11. [19]

    Kugele, A.; Pfeil, T.; Pfeiffer, M.; and Chicca, E. 2020. Efficient processing of spatio-temporal data streams with spiking neural networks. Frontiers in Neuroscience, 14: 512192

    2020

  12. [20]

    S.; Panda, P.; Srinivasan, G.; and Roy, K

    Lee, C.; Sarwar, S. S.; Panda, P.; Srinivasan, G.; and Roy, K. 2020. Enabling spike-based backpropagation for training deep neural network architectures. Frontiers in Neuroscience, 14: 497482

    2020

  13. [21]

    H.; Delbruck, T.; and Pfeiffer, M

    Lee, J. H.; Delbruck, T.; and Pfeiffer, M. 2016. Training deep spiking neural networks using backpropagation. Frontiers in Neuroscience, 10: 508

    2016

  14. [22]

    Liu, F.; Wang, Z.; Zhao, W.; Yang, N.; Chen, Y.; Huang, S.; Li, H.; Yang, T.; Pei, S.; Liang, X.; et al. 2024 a . Exploiting Temporal-Unrolled Parallelism for Energy-Efficient SNN Acceleration. IEEE Transactions on Parallel & Distributed Systems, (01): 1--16

    2024

  15. [23]

    Liu, Z.; Datta, G.; Li, A.; and Beerel, P. A. 2024 b . LMUFormer: Low Complexity Yet Powerful Spiking Model With Legendre Memory Units. In International Conference on Learning Representations

    2024

  16. [24]

    Logan, B.; et al. 2000. Mel frequency cepstral coefficients for music modeling. In Ismir, volume 270, 11

    2000

  17. [25]

    Maass, W. 1997. Networks of spiking neurons: the third generation of neural network models. Neural Networks, 10(9): 1659--1671

    1997

  18. [26]

    O.; Mostafa, H.; and Zenke, F

    Neftci, E. O.; Mostafa, H.; and Zenke, F. 2019. Surrogate gradient learning in spiking neural networks: Bringing the power of gradient-based optimization to spiking neural networks. IEEE Signal Processing Magazine, 36(6): 51--63

    2019

  19. [27]

    A.; and Roy, K

    Panda, P.; Aketi, S. A.; and Roy, K. 2020. Toward scalable, efficient, and accurate deep spiking neural networks with backward residual connections, stochastic softmax, and hybridization. Frontiers in Neuroscience, 14: 653

    2020

  20. [28]

    S.; Chan, W.; Zhang, Y.; Chiu, C.-C.; Zoph, B.; Cubuk, E

    Park, D. S.; Chan, W.; Zhang, Y.; Chiu, C.-C.; Zoph, B.; Cubuk, E. D.; and Le, Q. V. 2019. Specaugment: A simple data augmentation method for automatic speech recognition. arXiv preprint arXiv:1904.08779

    2019 arXiv

  21. [29]

    Rathi, N.; Chakraborty, I.; Kosta, A.; Sengupta, A.; Ankit, A.; Panda, P.; and Roy, K. 2023. Exploring neuromorphic computing based on spiking neural networks: Algorithms to hardware. ACM Computing Surveys, 55(12): 1--49

    2023

  22. [30]

    Rathi, N.; and Roy, K. 2021. 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, 34(6): 3174--3182

    2021

  23. [31]

    R.; and Panda, P

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

    2019

  24. [32]

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

    2019

  25. [33]

    J.; Schuster, M.; Jaitly, N.; Yang, Z.; Chen, Z.; Zhang, Y.; Wang, Y.; Skerrv-Ryan, R.; et al

    Shen, J.; Pang, R.; Weiss, R. J.; Schuster, M.; Jaitly, N.; Yang, Z.; Chen, Z.; Zhang, Y.; Wang, Y.; Skerrv-Ryan, R.; et al. 2018. Natural tts synthesis by conditioning wavenet on mel spectrogram predictions. In 2018 IEEE International Conference on Acoustics, Speech and Signa...

    2018

  26. [34]

    Shen, S.; Zhao, D.; Shen, G.; and Zeng, Y. 2024. TIM: An Efficient Temporal Interaction Module for Spiking Transformer. arXiv preprint arXiv:2401.11687

    2024 arXiv

  27. [35]

    M.; Shea, T.; Pacik-Nelson, N.; Gallo, E.; and Danielescu, A

    Stewart, K. M.; Shea, T.; Pacik-Nelson, N.; Gallo, E.; and Danielescu, A. 2023. Speech2spikes: Efficient audio encoding pipeline for real-time neuromorphic systems. In Proceedings of the 2023 Annual Neuro-Inspired Computational Elements Conference, 71--78

    2023

  28. [36]

    Su, J.; Ahmed, M.; Lu, Y.; Pan, S.; Bo, W.; and Liu, Y. 2024. Roformer: Enhanced transformer with rotary position embedding. Neurocomputing, 568: 127063

    2024

  29. [37]

    Su, Q.; Chou, Y.; Hu, Y.; Li, J.; Mei, S.; Zhang, Z.; and Li, G. 2023. Deep directly-trained spiking neural networks for object detection. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 6555--6565

    2023

  30. [38]

    Sun, P.; Chua, Y.; Devos, P.; and Botteldooren, D. 2023 a . Learnable axonal delay in spiking neural networks improves spoken word recognition. Frontiers in Neuroscience, 17: 1275944

    2023

  31. [39]

    Sun, P.; Eqlimi, E.; Chua, Y.; Devos, P.; and Botteldooren, D. 2023 b . Adaptive Axonal Delays in Feedforward Spiking Neural Networks for Accurate Spoken Word Recognition. In 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 1--5

    2023

  32. [40]

    N.; Kaiser, .; and Polosukhin, I

    Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A. N.; Kaiser, .; and Polosukhin, I. 2017. Attention is all you need. Advances in Neural Information Processing Systems, 30

    2017

  33. [41]

    H.; and Lim, M.-H

    Wang, S.; Cheng, T. H.; and Lim, M.-H. 2022. LTMD: learning improvement of spiking neural networks with learnable thresholding neurons and moderate dropout. Advances in Neural Information Processing Systems, 35: 28350--28362

    2022

  34. [42]

    Warden, P. 2018. Speech commands: A dataset for limited-vocabulary speech recognition. arXiv preprint arXiv:1804.03209

    2018 arXiv

  35. [43]

    Wu, X.; Zhao, Y.; Song, Y.; Jiang, Y.; Bai, Y.; Li, X.; Zhou, Y.; Yang, X.; and Hao, Q. 2023. Dynamic threshold integrate and fire neuron model for low latency spiking neural networks. Neurocomputing, 544: 126247

    2023

  36. [44]

    Xu, M.; Chen, X.; Guo, Y.; Wang, Y.; Qiu, D.; Du, X.; Cui, Y.; Wang, X.; and Xiong, J. 2023 a . Reconfigurable neuromorphic computing: Materials, devices, and integration. Advanced Materials, 35(51): 2301063

    2023

  37. [45]

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

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

    2023

  38. [46]

    Yang, H.; Lam, K.-Y.; Xiao, L.; Xiong, Z.; Hu, H.; Niyato, D.; and Vincent Poor, H. 2022. Lead federated neuromorphic learning for wireless edge artificial intelligence. Nature Communications, 13(1): 4269

    2022

  39. [47]

    Yang, Q.; Liu, Q.; and Li, H. 2022. Deep residual spiking neural network for keyword spotting in low-resource settings. In Interspeech, 3023--3027

    2022

  40. [48]

    Yang, S.; and Chen, B. 2023. SNIB: improving spike-based machine learning using nonlinear information bottleneck. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 53(12): 7852--7863

    2023

  41. [49]

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

    2021

  42. [50]

    Yao, M.; Hu, J.; Hu, T.; Xu, Y.; Zhou, Z.; Tian, Y.; XU, B.; and Li, G. 2024. Spike-driven Transformer V2: Meta Spiking Neural Network Architecture Inspiring the Design of Next-generation Neuromorphic Chips. In International Conference on Learning Representations

    2024

  43. [51]

    Zhang, H.; Zhou, C.; Yu, L.; Huang, L.; Ma, Z.; Fan, X.; Zhou, H.; and Tian, Y. 2024 a . SGLFormer: Spiking Global-Local-Fusion Transformer with High Performance. Frontiers in Neuroscience, 18: 1371290

    2024

  44. [52]

    Zhang, S.; Yang, Q.; Ma, C.; Wu, J.; Li, H.; and Tan, K. C. 2024 b . Tc-lif: A two-compartment spiking neuron model for long-term sequential modelling. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, 16838--16847

    2024

  45. [53]

    Zhou, C.; Zhang, H.; Yu, L.; Ye, Y.; Zhou, Z.; Huang, L.; Ma, Z.; Fan, X.; Zhou, H.; and Tian, Y. 2024. Direct training high-performance deep spiking neural networks: a review of theories and methods. Frontiers in Neuroscience, 18

    2024

  46. [54]

    Zhou, Z.; Zhu, Y.; He, C.; Wang, Y.; YAN, S.; Tian, Y.; and Yuan, L. 2023. Spikformer: When Spiking Neural Network Meets Transformer. In International Conference on Learning Representations

    2023

  47. [55]

    Zhu, R.-J.; Zhang, M.; Zhao, Q.; Deng, H.; Duan, Y.; and Deng, L.-J. 2024. Tcja-snn: Temporal-channel joint attention for spiking neural networks. IEEE Transactions on Neural Networks and Learning Systems, 1--14

    2024

  48. [56]

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

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all...

  49. [57]

    write newline

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

Pith tools

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