pith. sign in

arxiv: 2511.08558 · v2 · submitted 2025-11-11 · 💻 cs.AI

Hyperdimensional Decoding of Spiking Neural Networks

Cedrick Kinavuidi , Luca Peres , Oliver Rhodes This is my paper

Pith reviewed 2026-05-17 23:17 UTC · model grok-4.3

classification 💻 cs.AI
keywords spiking neural networkshyperdimensional computingdecoding methodsenergy efficiencyclassification accuracyneuromorphic computingunknown class detectionDVS datasets
0
0 comments X

The pith

A hyperdimensional computing decoder for spiking neural networks yields higher accuracy, lower latency, and reduced energy use while detecting untrained classes.

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

The paper introduces a decoding method for spiking neural networks that replaces traditional rate or latency coding with hyperdimensional computing. The central goal is to deliver high classification accuracy, strong noise tolerance, quick responses, and low power draw on neuromorphic hardware. A sympathetic reader would care because spiking networks promise efficient brain-like processing for sensors and edge devices, yet their output decoding has often limited real-world gains. The method also identifies classes never seen in training, reaching 100 percent detection on one benchmark set. If the results hold, the approach offers a practical alternative to existing decoding schemes across multiple datasets.

Core claim

The paper claims that an SNN decoded via hyperdimensional computing attains generally better classification accuracy, lower classification latency, and lower estimated energy consumption than analogous architectures decoded with existing methods, with energy reductions ranging from 1.24x to 3.67x on the DvsGesture dataset and 1.38x to 2.27x on the SL-Animals-DVS dataset, while also identifying 100 percent of samples from an unseen class on DvsGesture.

What carries the argument

The SNN-HDC decoding method, which encodes spiking network outputs into hyperdimensional vectors for classification instead of using rate or latency coding.

If this is right

  • The SNN-HDC model shows energy consumption reductions from 1.24x to 3.67x on the DvsGesture dataset.
  • The model identifies 100 percent of samples from an untrained class on the DvsGesture dataset.
  • Classification accuracy and latency improve across multiple literature test cases relative to standard decoding.
  • The approach maintains high noise robustness while lowering overall power draw.
  • The decoding method works as an alternative to both rate and latency coding.

Where Pith is reading between the lines

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

  • This decoding style may extend to other sensory modalities where spiking data must be interpreted under tight power budgets.
  • The ability to flag untrained classes could support incremental learning setups without retraining the entire network.
  • Hybrid hyperdimensional decoding might reduce the need for large output layers in neuromorphic classifiers.
  • If energy savings persist on physical chips, the method could influence hardware-software co-design for always-on edge sensors.

Load-bearing premise

The reported energy estimates accurately reflect real hardware deployment and that baseline comparisons use equivalent network sizes, training regimes, and hardware models.

What would settle it

Measure actual energy consumption and latency of the SNN-HDC model versus the baselines when both run on the same neuromorphic chip for identical tasks and input streams.

Figures

Figures reproduced from arXiv: 2511.08558 by Cedrick Kinavuidi, Luca Peres, Oliver Rhodes.

Figure 1
Figure 1. Figure 1: Two random binary hypervectors with 10 dimensions. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Number of classes that Binary Hypervectors and One-Hot Encoding can represent given a [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: A rate decoded SNN (A) and the SNN-HDC model (B). The rate decoded SNN has one [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: SNN-HDC results trained on the DvsGesture dataset [ [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Rate and latency decoded results trained on the DvsGesture dataset [ [PITH_FULL_IMAGE:figures/full_fig_p013_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: SNN-HDC, rate decoded and latency decoded layer activity. Numbers obtained over [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Average per layer firing rate of neurons in each model per sample of the DvsGesture [ [PITH_FULL_IMAGE:figures/full_fig_p015_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Accuracy obtained by the 10-class trained SNN-HDC when performing classification on [PITH_FULL_IMAGE:figures/full_fig_p016_8.png] view at source ↗
read the original abstract

This work presents a novel spiking neural network (SNN) decoding method, combining SNNs with Hyperdimensional computing (HDC). The goal is to create a decoding method with high accuracy, high noise robustness, low latency and low energy usage. Compared to analogous architectures decoded with existing approaches, the presented SNN-HDC model attains generally better classification accuracy, lower classification latency and lower estimated energy consumption on multiple test cases from literature. The SNN-HDC achieved estimated energy consumption reductions ranging from 1.24x to 3.67x on the DvsGesture dataset and from 1.38x to 2.27x on the SL-Animals-DVS dataset. The presented decoding method can also efficiently identify unknown classes it has not been trained on. In the DvsGesture dataset the SNN-HDC model can identify 100% of samples from an unseen/untrained class. Given the numerous benefits shown and discussed in this paper, this decoding method represents a very compelling alternative to both rate and latency decoding.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript proposes a novel decoding method for spiking neural networks that integrates hyperdimensional computing (HDC). It claims that the resulting SNN-HDC decoder achieves higher classification accuracy, lower latency, and lower estimated energy consumption than rate or latency decoding on literature benchmarks, with energy reductions of 1.24x–3.67x reported on DvsGesture and 1.38x–2.27x on SL-Animals-DVS. The method is also said to identify 100% of samples from an untrained class on DvsGesture.

Significance. If the energy model and baseline equivalence can be verified, the approach could supply a practical, noise-robust alternative for low-power neuromorphic decoding. The reported ability to flag unseen classes without retraining would be a useful property for open-set recognition tasks.

major comments (2)
  1. [Abstract] Abstract: the headline energy reductions (1.24x–3.67x on DvsGesture) are presented without any per-operation energy table, bit-width assumptions, memory-access model, or hardware parameters that were applied identically to both SNN-HDC and the rate/latency baselines. This omission makes the claimed savings unverifiable and is load-bearing for the central performance claim.
  2. [Abstract] Abstract: quantitative accuracy and latency gains are asserted without error bars, statistical tests, or a methods description of network sizes, training regimes, or evaluation protocols, preventing assessment of whether the reported improvements are robust or merely the result of unequal experimental conditions.
minor comments (1)
  1. [Abstract] The abstract refers to “multiple test cases from literature” but only names two datasets; a brief enumeration of all evaluated benchmarks would improve clarity.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments on our manuscript. We address each major comment point by point below. We have revised the abstract to improve self-containment and verifiability while preserving conciseness, and we refer to the detailed experimental and energy sections in the main text.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the headline energy reductions (1.24x–3.67x on DvsGesture) are presented without any per-operation energy table, bit-width assumptions, memory-access model, or hardware parameters that were applied identically to both SNN-HDC and the rate/latency baselines. This omission makes the claimed savings unverifiable and is load-bearing for the central performance claim.

    Authors: We agree that the abstract would benefit from additional context on the energy model to make the headline claims more immediately verifiable. The full manuscript presents the energy estimation methodology in Section 4, including per-operation costs, bit-width assumptions, and a consistent memory-access model applied identically to SNN-HDC and all baselines. We have revised the abstract to include a concise statement summarizing the uniform energy model and directing readers to the detailed breakdown and hardware parameters in the main text. revision: yes

  2. Referee: [Abstract] Abstract: quantitative accuracy and latency gains are asserted without error bars, statistical tests, or a methods description of network sizes, training regimes, or evaluation protocols, preventing assessment of whether the reported improvements are robust or merely the result of unequal experimental conditions.

    Authors: The manuscript reports accuracy and latency results with standard deviations across repeated trials and includes statistical comparisons in the results section, along with full descriptions of network architectures, training procedures, and evaluation protocols. We acknowledge that the abstract does not convey this information. We have revised the abstract to note that quantitative results are averaged over multiple runs with variability measures and to reference the methods section for network sizes, training regimes, and evaluation protocols. revision: yes

Circularity Check

0 steps flagged

No circularity; claims are empirical comparisons without self-referential derivations

full rationale

The paper introduces an SNN-HDC decoding method and reports performance gains (accuracy, latency, energy) via direct experimental comparisons on DvsGesture and SL-Animals-DVS datasets. No equations, fitted parameters renamed as predictions, or self-citation chains appear in the abstract or described content that would reduce any result to its own inputs by construction. Energy estimates are presented as outcomes of the evaluation rather than derived quantities forced by prior definitions within the work. The derivation chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on standard SNN training assumptions and HDC encoding choices whose details are not visible in the abstract; no new physical entities are introduced.

free parameters (1)
  • SNN and HDC hyperparameters
    Typical machine-learning model parameters tuned on the target datasets; exact values and tuning procedure unknown from abstract.
axioms (1)
  • domain assumption Standard assumptions of SNN training and HDC vector operations hold for the chosen datasets and hardware models.
    Invoked implicitly when claiming performance and energy advantages over baselines.

pith-pipeline@v0.9.0 · 5476 in / 1188 out tokens · 30657 ms · 2026-05-17T23:17:40.320875+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

56 extracted references · 56 canonical work pages · 4 internal anchors

  1. [1]

    Maass W 1997Neural Networks101659–1671 ISSN 0893-6080 URL https://www.sciencedirect.com/science/article/pii/S0893608097000117

    work page

  2. [2]

    Deng L, Wu Y, Hu X, Liang L, Ding Y, Li G, Zhao G, Li P and Xie Y 2020Neural Networks 121294–307 ISSN 0893-6080 URL https://www.sciencedirect.com/science/article/pii/S0893608019302667

    work page

  3. [3]

    Roy K, Jaiswal A and Panda P 2019Nature575607–617 ISSN 1476-4687 publisher: Nature Publishing Group URLhttps://www.nature.com/articles/s41586-019-1677-2

    work page

  4. [4]

    Yamazaki K, Vo-Ho V K, Bulsara D and Le N 2022Brain Sciences12863 ISSN 2076-3425 URLhttps://www.ncbi.nlm.nih.gov/pmc/articles/PMC9313413/

    work page 2076

  5. [5]

    Tavanaei A, Ghodrati M, Kheradpisheh S R, Masquelier T and Maida A 2019Neural Networks11147–63 ISSN 0893-6080 URL https://www.sciencedirect.com/science/article/pii/S0893608018303332

    work page

  6. [6]

    Attwell D and Laughlin S B 2001Journal of Cerebral Blood Flow & Metabolism211133–1145 ISSN 0271-678X publisher: SAGE Publications Ltd STM URL https://doi.org/10.1097/00004647-200110000-00001

    work page doi:10.1097/00004647-200110000-00001

  7. [7]

    Caccavella C, Paredes-Vall´ es F, Cannici M and Khacef L 2024 Low-power event-based face detection with asynchronous neuromorphic hardware2024 International Joint Conference on Neural Networks (IJCNN)pp 1–10 iSSN: 2161-4407 URL https://ieeexplore.ieee.org/abstract/document/10650843?signout=success

    work page arXiv 2024

  8. [8]

    Schuman C D, Potok T E, Patton R M, Birdwell J D, Dean M E, Rose G S and Plank J S 2017 A Survey of Neuromorphic Computing and Neural Networks in Hardware arXiv:1705.06963 URLhttp://arxiv.org/abs/1705.06963

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  9. [9]

    Davies M, Wild A, Orchard G, Sandamirskaya Y, Guerra G A F, Joshi P, Plank P and Risbud S R 2021Proceedings of the IEEE109911–934 ISSN 1558-2256 conference Name: Proceedings of the IEEE URLhttps://ieeexplore.ieee.org/document/9395703

    work page arXiv

  10. [10]

    Guo W, Fouda M E, Eltawil A M and Salama K N 2021Frontiers in Neuroscience15ISSN 1662-453X publisher: Frontiers URLhttps://www.frontiersin.org/journals/ neuroscience/articles/10.3389/fnins.2021.638474/full

    work page doi:10.3389/fnins.2021.638474/full 2021

  11. [11]

    Schuman C, Rizzo C, McDonald-Carmack J, Skuda N and Plank J 2022 Evaluating Encoding and Decoding Approaches for Spiking Neuromorphic SystemsProceedings of the International Conference on Neuromorphic Systems 2022(Knoxville TN USA: ACM) pp 1–9 ISBN 978-1-4503-9789-6 URLhttps://dl.acm.org/doi/10.1145/3546790.3546792

    work page doi:10.1145/3546790.3546792 2022

  12. [12]

    Davidson S and Furber S B 2021Frontiers in Neuroscience15651141 URL https://pmc.ncbi.nlm.nih.gov/articles/PMC8055931/

    work page

  13. [13]

    Caporale N and Dan Y 2008Annual Review of Neuroscience3125–46 ISSN 0147-006X, 1545-4126 publisher: Annual Reviews URLhttps: //www.annualreviews.org/content/journals/10.1146/annurev.neuro.31.060407.125639

    work page doi:10.1146/annurev.neuro.31.060407.125639

  14. [14]

    Barrett D G T, Den` eve S and Machens C K 2015 Optimal compensation for neuron death pages: 029512 Section: New Results URL https://www.biorxiv.org/content/10.1101/029512v1

    work page doi:10.1101/029512v1 2015

  15. [15]

    Jain A, Bansal R, Kumar A and Singh K D 2015International Journal of Applied and Basic Medical Research5124 URLhttps://pmc.ncbi.nlm.nih.gov/articles/PMC4456887/

    work page

  16. [16]

    Furber S 2012IEEE Spectrum4944–49 ISSN 1939-9340 conference Name: IEEE Spectrum URLhttps://ieeexplore.ieee.org/abstract/document/6247562

    work page arXiv 1939

  17. [17]

    Kanerva P 2009Cognitive Computation1139–159 ISSN 1866-9964 URL https://doi.org/10.1007/s12559-009-9009-8 20 IOP PublishingJournalvv(yyyy) aaaaaa C Kinavuidiet al

    work page doi:10.1007/s12559-009-9009-8

  18. [18]

    Gerstner W, Kistler W M, Naud R and Paninski L 2014Neuronal Dynamics: From Single Neurons to Networks and Models of Cognition1st ed (Cambridge University Press) ISBN 978-1-107-06083-8 URL https://www.cambridge.org/core/product/identifier/9781107447615/type/book

    work page arXiv

  19. [19]

    Eshraghian J K, Ward M, Neftci E, Wang X, Lenz G, Dwivedi G, Bennamoun M, Jeong D S and Lu W D 2023 Training Spiking Neural Networks Using Lessons From Deep Learning arXiv:2109.12894 [cs] URLhttp://arxiv.org/abs/2109.12894

    work page arXiv 2023

  20. [20]

    Auge D, Hille J, Mueller E and Knoll A 2021Neural Processing Letters534693–4710 ISSN 1573-773X URLhttps://doi.org/10.1007/s11063-021-10562-2

    work page doi:10.1007/s11063-021-10562-2

  21. [21]

    She X, Dash S and Mukhopadhyay S 2021 Sequence Approximation using Feedforward Spiking Neural Network for Spatiotemporal Learning: Theory and Optimization Methods URL https://openreview.net/forum?id=bp-LJ4y_XC

    work page 2021

  22. [22]

    Liu S, Leung V C H and Dragotti P L 2023Frontiers in Neuroscience17ISSN 1662-453X publisher: Frontiers URLhttps://www.frontiersin.org/journals/neuroscience/ articles/10.3389/fnins.2023.1266003/full

    work page doi:10.3389/fnins.2023.1266003/full 2023

  23. [23]

    Yin B, Guo Q, Corradi F and Bohte S 2022 Attentive Decision-making and Dynamic Resetting of Continual Running SRNNs for End-to-End Streaming Keyword Spotting Proceedings of the International Conference on Neuromorphic Systems 2022ICONS ’22 (New York, NY, USA: Association for Computing Machinery) pp 1–8 ISBN 978-1-4503-9789-6 URL https://dl.acm.org/doi/10....

    work page doi:10.1145/3546790.3546795 2022

  24. [24]

    Wu D, Jin G, Yu H, Yi X and Huang X 2025Frontiers in Neuroscience19ISSN 1662-453X publisher: Frontiers URLhttps://www.frontiersin.org/journals/neuroscience/ articles/10.3389/fnins.2025.1522788/full

    work page doi:10.3389/fnins.2025.1522788/full 2025

  25. [25]

    Klimo M, Luk´ aˇ c P and Tar´ abek P 2021Applied Sciences113563 ISSN 2076-3417 number: 8 Publisher: Multidisciplinary Digital Publishing Institute URL https://www.mdpi.com/2076-3417/11/8/3563

    work page 2076

  26. [26]

    Surv.55 130:1–130:40 ISSN 0360-0300 URLhttps://dl.acm.org/doi/10.1145/3538531

    Kleyko D, Rachkovskij D A, Osipov E and Rahimi A 2022ACM Comput. Surv.55 130:1–130:40 ISSN 0360-0300 URLhttps://dl.acm.org/doi/10.1145/3538531

    work page doi:10.1145/3538531

  27. [27]

    DeepSeek-AI, Liu A, Feng B, Xue B, Wang B, Wu B, Lu C, Zhao C, Deng C, Zhang C, Ruan C, Dai D, Guo D, Yang D, Chen D, Ji D, Li E, Lin F, Dai F, Luo F, Hao G, Chen G, Li G, Zhang H, Bao H, Xu H, Wang H, Zhang H, Ding H, Xin H, Gao H, Li H, Qu H, Cai J L, Liang J, Guo J, Ni J, Li J, Wang J, Chen J, Chen J, Yuan J, Qiu J, Li J, Song J, Dong K, Hu K, Gao K, G...

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  28. [28]

    Le˜ nero-Bardallo J A, Serrano-Gotarredona T and Linares-Barranco B 2011IEEE Journal of Solid-State Circuits461443–1455 ISSN 1558-173X conference Name: IEEE Journal of Solid-State Circuits URLhttps://ieeexplore.ieee.org/abstract/document/5746543

    work page arXiv

  29. [29]

    Liao F, Zhou F and Chai Y 2021Journal of Semiconductors42013105 ISSN 1674-4926 publisher: Chinese Institute of Electronics URL https://dx.doi.org/10.1088/1674-4926/42/1/013105 21 IOP PublishingJournalvv(yyyy) aaaaaa C Kinavuidiet al

    work page doi:10.1088/1674-4926/42/1/013105

  30. [30]

    Brandli C, Berner R, Yang M, Liu S C and Delbruck T 2014IEEE Journal of Solid-State Circuits492333–2341 ISSN 1558-173X URL https://ieeexplore.ieee.org/document/6889103

    work page arXiv

  31. [31]

    Chalich Y, Mallick A, Gupta B and Deen M J 2020PLoS ONE15e0232788 ISSN 1932-6203 URLhttps://www.ncbi.nlm.nih.gov/pmc/articles/PMC7209243/

    work page 1932

  32. [32]

    Abad G, Ersoy O, Picek S and Urbieta A 2024 Sneaky Spikes: Uncovering Stealthy Backdoor Attacks in Spiking Neural Networks with Neuromorphic DataProceedings 2024 Network and Distributed System Security SymposiumarXiv:2302.06279 [cs] URL http://arxiv.org/abs/2302.06279

    work page arXiv 2024

  33. [33]

    Apolinario M P E and Roy K 2024 S-TLLR: STDP-inspired Temporal Local Learning Rule for Spiking Neural Networks arXiv:2306.15220 version: 4 URL http://arxiv.org/abs/2306.15220

    work page arXiv 2024

  34. [34]

    Sun H, Cai W, Yang B, Cui Y, Xia Y, Yao D and Guo D 2023 A Synapse-Threshold Synergistic Learning Approach for Spiking Neural Networks arXiv:2206.06129 [cs] version: 3 URLhttp://arxiv.org/abs/2206.06129

    work page arXiv 2023

  35. [35]

    Venkatesh S, Marinescu R and Eshraghian J K 2024 SQUAT: Stateful Quantization-Aware Training in Recurrent Spiking Neural Networks2024 Neuro Inspired Computational Elements Conference (NICE)pp 1–10 URL https://ieeexplore.ieee.org/abstract/document/10549198/figures#figures

    work page arXiv 2024

  36. [36]

    Amir A, Taba B, Berg D, Melano T, McKinstry J, Di Nolfo C, Nayak T, Andreopoulos A, Garreau G, Mendoza M, Kusnitz J, Debole M, Esser S, Delbruck T, Flickner M and Modha D 2017 A Low Power, Fully Event-Based Gesture Recognition System2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)pp 7388–7397 iSSN: 1063-6919 URL https://ieeexplore.ie...

    work page arXiv 2017

  37. [37]

    Zou Z, Alimohamadi H, Zakeri A, Imani F, Kim Y, Najafi M H and Imani M 2022Scientific Reports127641 ISSN 2045-2322 URL https://www.nature.com/articles/s41598-022-11073-3

    work page 2045

  38. [38]

    Kaiser J, Mostafa H and Neftci E 2020Frontiers in Neuroscience14ISSN 1662-453X URL https://www.frontiersin.org/articles/10.3389/fnins.2020.00424

    work page doi:10.3389/fnins.2020.00424 2020

  39. [39]

    Deng L 2012IEEE Signal Processing Magazine29141–142 ISSN 1558-0792 conference Name: IEEE Signal Processing Magazine URL https://ieeexplore.ieee.org/abstract/document/6296535

    work page arXiv

  40. [40]

    Morris J, Lui H W, Stewart K, Khaleghi B, Thomas A, Marback T, Aksanli B, Neftci E and Rosing T 2022 HyperSpike: HyperDimensional Computing for More Efficient and Robust Spiking Neural Networks2022 Design, Automation & Test in Europe Conference & Exhibition (DATE)pp 664–669 iSSN: 1558-1101 URL https://ieeexplore.ieee.org/document/9774644

    work page arXiv 2022

  41. [41]

    Orchard G, Jayawant A, Cohen G K and Thakor N 2015Frontiers in Neuroscience9ISSN 1662-453X URLhttps://www.frontiersin.org/articles/10.3389/fnins.2015.00437

    work page doi:10.3389/fnins.2015.00437 2015

  42. [42]

    Bi Y, Chadha A, Abbas A, Bourtsoulatze E and Andreopoulos Y 2019 Graph-Based Object Classification for Neuromorphic Vision Sensing2019 IEEE/CVF International Conference on Computer Vision (ICCV)pp 491–501 iSSN: 2380-7504 URL https://ieeexplore.ieee.org/document/9010397

    work page arXiv 2019

  43. [43]

    Galvez R L, Dadios E P, Bandala A A and Vicerra R R P 2019 YOLO-based Threat Object Detection in X-ray Images2019 IEEE 11th International Conference on Humanoid, Nanotechnology, Information Technology, Communication and Control, Environment, and Management ( HNICEM )pp 1–5 URLhttps://ieeexplore.ieee.org/document/9073599

    work page arXiv 2019

  44. [44]

    Li X, Wang W, Xue F and Song Y 2018Physica A: Statistical Mechanics and its Applications 491716–728 ISSN 0378-4371 URL https://www.sciencedirect.com/science/article/pii/S0378437117307896 22 IOP PublishingJournalvv(yyyy) aaaaaa C Kinavuidiet al

    work page

  45. [45]

    Horowitz M 2014 1.1 Computing’s energy problem (and what we can do about it)2014 IEEE International Solid-State Circuits Conference Digest of Technical Papers (ISSCC)pp 10–14 iSSN: 2376-8606 URLhttps://ieeexplore.ieee.org/document/6757323

    work page arXiv 2014

  46. [46]

    URLhttps://zenodo.org/records/5079802

    Lenz G, Chaney K, Shrestha S B, Oubari O, Picaud S and Zarrella G 2021 Tonic: event-based datasets and transformations. URLhttps://zenodo.org/records/5079802

    work page arXiv 2021

  47. [47]

    Vasudevan A, Negri P, Di Ielsi C, Linares-Barranco B and Serrano-Gotarredona T 2022 Pattern Analysis and Applications25505–520 ISSN 1433-755X URL https://doi.org/10.1007/s10044-021-01011-w

    work page doi:10.1007/s10044-021-01011-w 2022

  48. [48]

    Sun M, Zhang D, Ge Z, Wang J, Li J, Fang Z and Xu R 2024 EventRPG: Event Data Augmentation with Relevance Propagation Guidance arXiv:2403.09274 [cs] URL http://arxiv.org/abs/2403.09274

    work page arXiv 2024

  49. [49]

    A sparse-aware solution for efficient event data processing arXiv:2204.03355 [cs] URLhttp://arxiv.org/abs/2204.03355

    Sabater A, Montesano L and Murillo A C 2022 Event Transformer. A sparse-aware solution for efficient event data processing arXiv:2204.03355 [cs] URLhttp://arxiv.org/abs/2204.03355

    work page arXiv 2022

  50. [50]

    Furber S B, Lester D R, Plana L A, Garside J D, Painkras E, Temple S and Brown A D 2013 IEEE Transactions on Computers622454–2467 ISSN 1557-9956 URL https://ieeexplore.ieee.org/document/6226357

    work page arXiv 2013

  51. [51]

    Kingma D P and Ba J 2017 Adam: A Method for Stochastic Optimization arXiv:1412.6980 [cs] URLhttp://arxiv.org/abs/1412.6980

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  52. [52]

    Ansel J, Yang E, He H, Gimelshein N, Jain A, Voznesensky M, Bao B, Bell P, Berard D, Burovski E, Chauhan G, Chourdia A, Constable W, Desmaison A, DeVito Z, Ellison E, Feng W, Gong J, Gschwind M, Hirsh B, Huang S, Kalambarkar K, Kirsch L, Lazos M, Lezcano M, Liang Y, Liang J, Lu Y, Luk C K, Maher B, Pan Y, Puhrsch C, Reso M, Saroufim M, Siraichi M Y, Suk H...

    work page doi:10.1145/3620665.3640366 2024

  53. [53]

    Merolla P A, Arthur J V, Alvarez-Icaza R, Cassidy A S, Sawada J, Akopyan F, Jackson B L, Imam N, Guo C, Nakamura Y, Brezzo B, Vo I, Esser S K, Appuswamy R, Taba B, Amir A, Flickner M D, Risk W P, Manohar R and Modha D S 2014Science345668–673 publisher: American Association for the Advancement of Science URL https://www.science.org/doi/10.1126/science.1254642

    work page doi:10.1126/science.1254642

  54. [54]

    Li Y, Geller T, Kim Y and Panda P 2023 SEENN: Towards Temporal Spiking Early-Exit Neural Networks arXiv:2304.01230 [cs] URLhttp://arxiv.org/abs/2304.01230

    work page arXiv 2023

  55. [55]

    Liu C, Qi X, Lam E Y and Wong N 2022IEEE Access1055638–55649 ISSN 2169-3536 conference Name: IEEE Access URLhttps://ieeexplore.ieee.org/document/9781403

    work page arXiv

  56. [56]

    Shrestha S B and Orchard G 2018 SLAYER: Spike Layer Error Reassignment in Time arXiv:1810.08646 [cs] URLhttp://arxiv.org/abs/1810.08646 23

    work page internal anchor Pith review Pith/arXiv arXiv 2018