Pith. sign in

REVIEW 4 major objections 6 minor 49 references

Integer Binary-Range Alignment Neuron for Spiking Neural Networks

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Spiking neurons that fire integer codes, converted to binary spikes at inference, match or beat the same-architecture ANN on ImageNet (74.19% vs 73.84%) at a computed 6.3× lower energy cost.

desk verdict The binary-encoding neuron idea is new and the results are strong, but a bit-indexing typo in Eq. (6) and flaky cost accounting mean the paper is not reproducible as printed. read the letter →

arxiv 2506.05679 v1 pith:2IDRDF5S submitted 2025-06-06 cs.NE cs.CV

Binghao Ye , Wenjuan Li , Dong Wang , Man Yao , Bing Li , Weiming Hu , Dong Liang , Kun Shang This is my paper
classification cs.NEcs.CV
keywords spikingneuralnetworksintegerfiringbinaryencodingrangealignmentinformationexpressioncapacitydirecttrainingANN-to-SNNconversionenergy-efficientinference
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 sets out to close the accuracy gap between spiking neural networks and ordinary artificial neural networks without giving up spike-driven, accumulate-only inference. Its proposal, the IBRA-LIF neuron, fires positive integer values during training and then, at inference, decomposes each integer into binary bit-planes that behave like parallel 0/1 spike streams. Because convolution is linear, this decomposition is exact, and a single timestep carries exponentially more information than a single bit. A range-alignment scaling factor prevents the neurons from only ever firing small values, which would leave most of that capacity unused. The reported outcome is that directly trained SNNs of this kind beat same-architecture ANNs on ImageNet (74.19% versus 73.84%) and on CIFAR100, at a computed 6.3× lower energy cost.

What carries the argument

The load-bearing object is an exact identity for linear layers: $W \sum_{b} 2^{b} O^{b} = \sum_{b} 2^{b} W O^{b}$. It converts one integer-valued forward pass into $B$ parallel 0/1 spike streams whose weighted sum is bit-identical to the original, turning MAC-heavy convolutions into $B$ accumulate-only operations and giving exponential capacity growth for linear energy growth. Two supporting mechanisms carry the training side: the range-alignment factor $N$ that scales the membrane potential before rounding and clipping (with the division folded into the weights as $W^{N} = W/N$ at inference), and the straight-through surrogate gradient $\varphi(V)=1$ for $0 \le V \le D$ that lets backpropagation pass through the non-differentiable integer firing.

What would settle it

Run the trained ImageNet network and record, layer by layer, how often each of the 9 binary digits of $O^t_l \times N$ is active. If the top bit-planes stay silent in most layers, the exponential capacity expansion is not actually being used, and a different explanation (such as plain integer quantization or range normalization) would need to account for the accuracy gain.

Watch

Extended reading notes

Core claim

IBRA-LIF, on the paper's own terms, is a spiking neuron that replaces the binary firing step with an integer emission. During training the neuron outputs $O^{t} = \mathrm{Clip}(\mathrm{round}(V_{\mathrm{pre}}^{t} \times N),\, 0,\, DN)/N$, which keeps the information loss of direct training low; during inference the integer is rewritten in binary as $O^{t}_{l} = \sum_{b=1}^{B} 2^{b}\, O^{t,b}_{l}$ with $O^{t,b}_{l} = (O^{t}_{l} \gg b)\ \&\ 1$, so the $B$ bit-planes are ordinary 0/1 spike streams. Because convolution is linear, the next layer computes $\sum_{b=1}^{B} 2^{b}\, W^{N}_{l+1} O^{t,b}_{l}$ with a re-parameterized weight $W^{N} = W/N$: pure accumulate operations on spikes. The authors claim that this yields an exponentially larger theoretical expression capacity per unit energy than unary integer firing, and they report directly trained IBRA-LIF reaching 74.19% top-1 on ImageNet with ResNet34 — 0.35% above the same architecture's ANN — and 49.2% mAP@50:95 on COCO, at a computed 6.3× energy saving over the ANN, plus near-lossless ANN-to-SNN conversion at 73.04%.

Load-bearing premise

The argument assumes that one global scaling factor, N=100, tuned once on a small dataset, correctly matches the neuron's voltage range to its integer output range in every layer and every task; if the ranges drift in deeper or wider networks, the high part of the integer code never fires and the reported gains disappear.

Editorial extensions

If this is right

  • Directly trained spiking networks can match or beat the same-architecture ANN: +0.35% on ImageNet (74.19% vs 73.84%) and +0.11% on CIFAR100 (81.67% vs 81.56%).
  • ANN-to-SNN conversion becomes near-lossless and fast: 73.04% top-1 on ImageNet versus 71.12% for the best prior conversion baseline, which needed 64 timesteps.
  • Object detection improves to 66.2% mAP@50 and 49.2% mAP@50:95 on COCO, surpassing same-backbone prior SNN detectors and a similarly sized YOLOv5.
  • Estimated energy stays low: the N=100 model is computed at 6.3× the ANN's efficiency on the CIFAR10 energy analysis, and the N=10 model uses 63% of LIF's energy while improving accuracy.
  • Effective latency multiplies: a network run for T timesteps executes as T×B bit-plane steps, each an accumulate-only 0/1 convolution, so expression capacity grows exponentially with B at linear energy cost.

Reading between the lines

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

  • A per-layer or learned scaling factor $N$ is a natural next step the paper does not run; layer-wise activation-range statistics could set $N_l$ where the single global $N=100$ leaves bit-planes silent, likely recovering more capacity in deep layers.
  • Because the binary decomposition relies only on linearity, the integer-plus-bit-plane trick transfers to other spike-driven architectures — Transformers, segmentation heads, or event-stream models — where it acts as an exact refinement of the temporal code rather than an approximation.
  • Ablating the training/inference asymmetry would localize the source of the gain: training forwards raw integers while inference uses binary reconstruction, and swapping one for the other would quantify how much of the improvement is true capacity expansion versus quantization side effects.
  • The paper itself notes that current hardware confines IBRA-LIF to synchronous neuromorphic chips; mapping the bit-plane expansion onto asynchronous hardware, where spike timing replaces a global clock, is an open engineering step that would decide how much of the theoretical energy saving survives in practice.
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

4 major / 6 minor

Summary. The paper proposes IBRA-LIF, a spiking neuron that trains with integer-valued activations produced by rounding and clipping a scaled membrane potential, and at inference converts the rounded integer into a binary spike expansion, using the linearity of convolution to perform accumulate-only operations with re-parameterized weights. A range alignment factor N is introduced to scale the membrane potential into the neuron's output range. The authors report state-of-the-art results on CIFAR10/100, ImageNet, CIFAR10-DVS, and COCO, including 74.19% top-1 accuracy on ImageNet with ResNet34, 49.2% mAP@50:95 on COCO, and 6.3x energy efficiency relative to the ANN baseline, together with near-lossless ANN-to-SNN conversion at one timestep.

Significance. The core idea is attractive: binary decomposition of integer activations can multiply the effective temporal information capacity while preserving spike-driven, accumulate-only inference, provided the binary reconstruction is exact. The linearity argument in Eq. (8) and the weight re-parameterization in Eqs. (13)-(15) are mathematically sound, and the reported empirical gains over existing SNN methods are large and consistent across classification and detection. If the implementation matches the corrected version of the method, this is a practically valuable contribution to high-performance low-energy SNNs. However, the manuscript as submitted contains an off-by-one error in the central binary reconstruction equation, leaves the neuron dynamics under-specified, and provides no code or statistical validation, so the paper is not currently reproducible from the text.

major comments (4)
  1. [Sec. 3.2, Eq. (6) and Eq. (12)] The binary expansion has an off-by-one indexing error. With O^{t,b}_l = (O^t_l >> b) & 1 and weights 2^b for b = 1..B, the reconstructed value is sum_{b=1}^B 2^b bit_b = 2*floor(O/2), not O. For example, O=3 reconstructs to 2 and O=5 to 4. Because Eq. (10) produces activations at multiples of 1/N, odd multiples of 1/N are unreachable at inference. This invalidates the exact reconstruction that justifies the exponential capacity gain and the AC-based inference in Eqs. (8)-(9). Please correct the summation to b=0..B-1 with weights 2^b, or b=1..B with weights 2^{b-1}, adjust the definition of B accordingly, and confirm that the reported experiments used the corrected convention.
  2. [Sec. 4.2, Table 1; Secs. 4.3-4.5] The hyperparameters N=100 and D=5.11 are selected on CIFAR10 and then transferred without modification to CIFAR100, ImageNet, CIFAR10-DVS, and COCO. The paper provides no per-layer membrane-potential range statistics and no theoretical bound relating the global factor N to the distribution of V_pre. If the activation range differs by architecture or dataset, the high binary bits will remain unused and the reported capacity gains will disappear. Please provide per-layer range statistics or a sensitivity analysis (for example N=10 and N=1000) on at least ImageNet and COCO.
  3. [Secs. 4.3-4.5, Tables 2, 5, and 6] Several headline claims rest on differences that cannot be assessed without variance information, such as +0.35% over the ANN on ImageNet in Table 5 and +0.11% over the ANN on CIFAR100 ResNet19 in Table 2. All results are single runs with no error bars or multiple seeds, and no code is released. The need for verification is amplified by the inconsistency between Eq. (6) and the reported results. Please report mean and standard deviation over at least three seeds for the smaller datasets and clarify the code-release plan.
  4. [Secs. 3.1-3.3, Eqs. (1)-(4), (10), and (16)] The manuscript never specifies the membrane-potential dynamics for IB-LIF or IBRA-LIF after the output becomes an integer, or an integer divided by N. Equation (3) is the reset rule for a binary LIF; for integer outputs it is unclear whether the reset subtracts O, O*N, or something else. Similarly, Eq. (16) defines a surrogate gradient on V but does not state how the round operation and the 1/N scaling in Eq. (10) are treated in the backward pass. These omissions leave the method under-specified and leave the training/inference consistency of the quantization step unanalyzed. Please add the full charging, firing, and resetting equations and the exact backward rule.
minor comments (6)
  1. [Sec. 4.2, Table 1] The table is difficult to read: the values in the D×N column appear to be the exponent B rather than the product, and entries such as "102" and "1002" are unreadable. Please regenerate the table with clear column headers and superscripts.
  2. [Abstract and Sec. 4.5, Table 6] The abstract reports 49.1% mAP@50:95 on COCO, while Table 6 reports 49.2% for the largest IBRA-LIF model; please reconcile the two numbers.
  3. [Secs. 4.3-4.5, Tables 2-6] The tables report T×D, but the actual number of inference timesteps is T×B, where B is the number of bits. Reporting T×B as well would make the comparisons with prior work more transparent.
  4. [Sec. 3.3, Eq. (10)] The notation D_N is introduced but later written as "DN" without consistency; please define D_N once and use it uniformly.
  5. [Throughout] There are several typos and unfinished items: "Emergy" in Table 4, "conprised" in Sec. 3.3, and reference [41] contains the placeholder "Accessed: insert date here."
  6. [Sec. 4.2, Table 1] The accuracy column contains leading minus signs, for example "−196.89%", which should be removed.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: binary expansion and range alignment are definitional/mathematical, and no fitted quantity is relabeled as a prediction.

full rationale

The paper's derivation chain is self-contained. Equations (6)-(9) are the binary expansion identity applied to integer neuron outputs; the claimed exponential reduction in spike count follows directly from the definition of binary representation, not from a fitted or data-dependent quantity. Equation (10) defines the range-alignment scaling N; its value (N=100, D=5.11) is chosen by a CIFAR10 ablation and then transferred to other tasks, which is ordinary hyperparameter selection, not a circular 'prediction' of the downstream accuracies. The energy comparison is an accounting exercise over the same binary representation; it is not a statistical prediction forced by the inputs. The citations to I-LIF/SpikeYOLO [30] and other prior work are descriptive and comparative; although author overlap exists, no load-bearing premise is justified solely by a self-citation, and no uniqueness theorem is imported. One correctness caveat, not circularity: Eq. (6) as printed sums b=1..B with weights 2^b, which drops bit 0 and over-weights the remaining bits, so the reconstruction formula does not equal O as written; this is a reproducibility/indexing issue that should be corrected but does not make the derivation circular. Overall circularity score: 0.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central claim rests on two fitted hyperparameters (D and N) and on three unproved background assumptions: convolution linearity (standard), surrogate gradient utility (domain assumption), and the transferability of a single global scaling factor N (ad hoc). No new physical entities are introduced.

free parameters (2)
  • D (maximum integer output) = 5.11 (optimal on CIFAR10 ablation; used in most experiments)
    Controls the maximum integer activation. Chosen by ablation on CIFAR10 with ResNet18; D=5.11 with N=100 gives D×N=511 (9 bits). The value is not derived from theory.
  • N (range alignment scaling) = 100 (optimal on CIFAR10; used for all later tasks)
    Multiplies membrane potential before firing and divides outputs. Selected via CIFAR10 ablation; assumed to transfer across layers and datasets with no per-layer or per-task adaptation.
assumptions (3)
  • standard math Convolution is linear, so binary bit decomposition can be moved outside the convolution (Eq. 8-9)
    Used to reorder the sum over bits and the convolution weight multiplication; mathematically valid.
  • domain assumption The surrogate gradient (STE) in Eq. (16), with a constant derivative in [0,D], yields useful gradients for training
    Standard in SNN direct training, but not analyzed for the specific integer rounding and N scaling; clipping at D may cause vanishing gradients.
  • ad hoc to paper A single global N aligns the membrane potential ranges in all layers
    RA's effectiveness depends on V_pre×N spanning [0, DN] in every layer; the paper gives no empirical distribution plots or theoretical derivation, only the CIFAR10 ablation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Integer Binary-Range Alignment Neuron for Spiking Neural Networks." pith.science (2026). https://pith.science/paper/2IDRDF5S

@misc{pith2026250605679,
  author       = {Pith},
  title        = {Pith review of: Integer Binary-Range Alignment Neuron for Spiking Neural Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2IDRDF5S}},
  note         = {Machine review of arXiv:2506.05679}
}
abstract

Spiking Neural Networks (SNNs) are noted for their brain-like computation and energy efficiency, but their performance lags behind Artificial Neural Networks (ANNs) in tasks like image classification and object detection due to the limited representational capacity. To address this, we propose a novel spiking neuron, Integer Binary-Range Alignment Leaky Integrate-and-Fire to exponentially expand the information expression capacity of spiking neurons with only a slight energy increase. This is achieved through Integer Binary Leaky Integrate-and-Fire and range alignment strategy. The Integer Binary Leaky Integrate-and-Fire allows integer value activation during training and maintains spike-driven dynamics with binary conversion expands virtual timesteps during inference. The range alignment strategy is designed to solve the spike activation limitation problem where neurons fail to activate high integer values. Experiments show our method outperforms previous SNNs, achieving 74.19% accuracy on ImageNet and 66.2% mAP@50 and 49.1% mAP@50:95 on COCO, surpassing previous bests with the same architecture by +3.45% and +1.6% and +1.8%, respectively. Notably, our SNNs match or exceed ANNs' performance with the same architecture, and the energy efficiency is improved by 6.3${\times}$.

Figures

Figures reproduced from arXiv: 2506.05679 by the authors.

Figure 1
Figure 1. a) demonstrates the differences between IB-LIF during the training and inference phases. We implement binary conversion [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The similarities and differences between IB-LIF and I [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. The spike activation limitation problem. when we set [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Comparison of LIF, I-LIF, and IBRA-LIF. LIF emits only 0/1 spikes during both training and inference, leading to significant [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

49 extracted references · 43 canonical work pages

  1. [1]

    Enhancing train- ing of spiking neural network with stochastic latency

    Srinivas Anumasa, Bhaskar Mukhoty, Velibor Bojkovic, Giulia De Masi, Huan Xiong, and Bin Gu. Enhancing train- ing of spiking neural network with stochastic latency. InPro- ceedings of the AAAI Conference on Artificial Intelligence, pages 10900–10908, 2024. 7

    work page 2024

  2. [2]

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

    Tong Bu, Wei Fang, Jianhao Ding, PENGLIN DAI, Zhaofei Yu, and Tiejun Huang. Optimal ann-snn conversion for high- accuracy and ultra-low-latency spiking neural networks. InInternational Conference on Learning Representations,

    work page

  3. [3]

    Spiking deep convolutional neural networks for energy-efficient ob- ject recognition.International Journal of Computer Vision, 113:54–66, 2015

    Yongqiang Cao, Yang Chen, and Deepak Khosla. Spiking deep convolutional neural networks for energy-efficient ob- ject recognition.International Journal of Computer Vision, 113:54–66, 2015. 2

    work page 2015

  4. [4]

    End-to- end object detection with transformers

    Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas Usunier, Alexander Kirillov, and Sergey Zagoruyko. End-to- end object detection with transformers. InEuropean confer- ence on computer vision, pages 213–229. Springer, 2020. 8

    work page 2020

  5. [5]

    Imagenet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In2009 IEEE conference on computer vision and pattern recognition, pages 248–255. Ieee, 2009. 5

    2009

  6. [6]

    Deep residual learning in spiking neural networks.Advances in Neural Information Processing Systems, 34:21056–21069, 2021

    Wei Fang, Zhaofei Yu, Yanqi Chen, Tiejun Huang, Timoth´ee Masquelier, and Yonghong Tian. Deep residual learning in spiking neural networks.Advances in Neural Information Processing Systems, 34:21056–21069, 2021. 8

    work page 2021

  7. [7]

    Incorporating learnable membrane time constant to enhance learning of spiking neu- ral networks

    Wei Fang, Zhaofei Yu, Yanqi Chen, Timoth ´ee Masquelier, Tiejun Huang, and Yonghong Tian. Incorporating learnable membrane time constant to enhance learning of spiking neu- ral networks. InProceedings of the IEEE/CVF international conference on computer vision, pages 2661–2671, 2021. 6

    work page 2021

  8. [8]

    Im-loss: information maximization loss for spiking neural networks.Advances in Neural Information Processing Systems, 35:156–166, 2022

    Yufei Guo, Yuanpei Chen, Liwen Zhang, Xiaode Liu, Yin- glei Wang, Xuhui Huang, and Zhe Ma. Im-loss: information maximization loss for spiking neural networks.Advances in Neural Information Processing Systems, 35:156–166, 2022. 2

    work page 2022

Show all 49 references
  1. [9]

    Reducing information loss for spiking neural networks

    Yufei Guo, Yuanpei Chen, Liwen Zhang, YingLei Wang, Xi- aode Liu, Xinyi Tong, Yuanyuan Ou, Xuhui Huang, and Zhe Ma. Reducing information loss for spiking neural networks. InEuropean Conference on Computer Vision, pages 36–52. Springer, 2022. 2, 8

    2022

  2. [10]

    Real spike: Learning real-valued spikes for spiking neural net- works

    Yufei Guo, Liwen Zhang, Yuanpei Chen, Xinyi Tong, Xi- aode Liu, YingLei Wang, Xuhui Huang, and Zhe Ma. Real spike: Learning real-valued spikes for spiking neural net- works. InEuropean Conference on Computer Vision, pages 52–68. Springer, 2022. 1, 2, 6, 7

    2022

  3. [11]

    Rmp- loss: Regularizing membrane potential distribution for spik- ing neural networks

    Yufei Guo, Xiaode Liu, Yuanpei Chen, Liwen Zhang, Wei- hang Peng, Yuhan Zhang, Xuhui Huang, and Zhe Ma. Rmp- loss: Regularizing membrane potential distribution for spik- ing neural networks. InProceedings of the IEEE/CVF In- ternational Conference on Computer Vision, pages 173...

    2023

  4. [12]

    Joint a-snn: Joint training of artificial and spiking neural networks via self- distillation and weight factorization.Pattern Recognition, 142:109639, 2023

    Yufei Guo, Weihang Peng, Yuanpei Chen, Liwen Zhang, Xi- aode Liu, Xuhui Huang, and Zhe Ma. Joint a-snn: Joint training of artificial and spiking neural networks via self- distillation and weight factorization.Pattern Recognition, 142:109639, 2023. 6

    2023

  5. [13]

    Mem- brane potential batch normalization for spiking neural net- works

    Yufei Guo, Yuhan Zhang, Yuanpei Chen, Weihang Peng, Xi- aode Liu, Liwen Zhang, Xuhui Huang, and Zhe Ma. Mem- brane potential batch normalization for spiking neural net- works. InProceedings of the IEEE/CVF International Con- ference on Computer Vision, pages 19420–19430, 2023. 2

    2023

  6. [14]

    Ternary spike: Learning ternary spikes for spiking neural networks

    Yufei Guo, Yuanpei Chen, Xiaode Liu, Weihang Peng, Yuhan Zhang, Xuhui Huang, and Zhe Ma. Ternary spike: Learning ternary spikes for spiking neural networks. InPro- ceedings of the AAAI Conference on Artificial Intelligence, pages 12244–12252, 2024. 1, 2, 7, 8

    2024

  7. [15]

    Deep spiking neural network: Energy efficiency through time based coding

    Bing Han and Kaushik Roy. Deep spiking neural network: Energy efficiency through time based coding. InEuropean Conference on Computer Vision, pages 388–404. Springer,

  8. [16]

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

    Bing Han, Gopalakrishnan Srinivasan, and Kaushik Roy. Rmp-snn: Residual membrane potential neuron for enabling deeper high-accuracy and low-latency spiking neural net- work. InProceedings of the IEEE/CVF conference on com- puter vision and pattern recognition, pages 13558–13567,

  9. [17]

    Reducing ann-snn conversion error through residual membrane potential

    Zecheng Hao, Tong Bu, Jianhao Ding, Tiejun Huang, and Zhaofei Yu. Reducing ann-snn conversion error through residual membrane potential. InProceedings of the AAAI Conference on Artificial Intelligence, pages 11–21, 2023. 2, 8

    2023

  10. [18]

    A progressive training framework for spiking neural networks with learnable multi-hierarchical model

    Zecheng Hao, Xinyu Shi, Zihan Huang, Tong Bu, Zhaofei Yu, and Tiejun Huang. A progressive training framework for spiking neural networks with learnable multi-hierarchical model. InThe Twelfth International Conference on Learning Representations, 2023. 7

    2023

  11. [19]

    Batch normalization: Accelerating deep network training by reducing internal co- variate shift

    Sergey Ioffe and Christian Szegedy. Batch normalization: Accelerating deep network training by reducing internal co- variate shift. InInternational conference on machine learn- ing, pages 448–456. pmlr, 2015. 5

    2015

  12. [20]

    A unified optimization framework of 9 ann-snn conversion: towards optimal mapping from activa- tion values to firing rates

    Haiyan Jiang, Srinivas Anumasa, Giulia De Masi, Huan Xiong, and Bin Gu. A unified optimization framework of 9 ann-snn conversion: towards optimal mapping from activa- tion values to firing rates. InInternational Conference on Machine Learning, pages 14945–14974. PMLR, 2023. 7, 8

    2023

  13. [21]

    Tab: Temporal accumulated batch normal- ization in spiking neural networks

    Haiyan Jiang, Vincent Zoonekynd, Giulia De Masi, Bin Gu, and Huan Xiong. Tab: Temporal accumulated batch normal- ization in spiking neural networks. InThe Twelfth Interna- tional Conference on Learning Representations, 2024. 7, 8

    2024

  14. [22]

    Towards fast and accurate object detection in bio-inspired spiking neural networks through bayesian optimization.IEEE Access, 9:2633–2643, 2020

    Seijoon Kim, Seongsik Park, Byunggook Na, Jongwan Kim, and Sungroh Yoon. Towards fast and accurate object detection in bio-inspired spiking neural networks through bayesian optimization.IEEE Access, 9:2633–2643, 2020. 8

    2020

  15. [23]

    Spiking-yolo: spiking neural network for energy- efficient object detection

    Seijoon Kim, Seongsik Park, Byunggook Na, and Sungroh Yoon. Spiking-yolo: spiking neural network for energy- efficient object detection. InProceedings of the AAAI con- ference on artificial intelligence, pages 11270–11277, 2020. 8

    2020

  16. [24]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky, Geoffrey Hinton, et al. Learning multiple layers of features from tiny images. 2009. 5

    2009

  17. [25]

    Cifar10-dvs: an event-stream dataset for ob- ject classification.Frontiers in neuroscience, 11:309, 2017

    Hongmin Li, Hanchao Liu, Xiangyang Ji, Guoqi Li, and Luping Shi. Cifar10-dvs: an event-stream dataset for ob- ject classification.Frontiers in neuroscience, 11:309, 2017. 5

    2017

  18. [26]

    Efficient and accurate conversion of spiking neural network with burst spikes

    Yang Li and Yi Zeng. Efficient and accurate conversion of spiking neural network with burst spikes. InProceedings of the Thirty-First International Joint Conference on Artificial Intelligence, IJCAI-22, pages 2485–2491. International Joint Conferences on Artificial Intelligenc...

    2022

  19. [27]

    A free lunch from ann: Towards efficient, accurate spiking neural networks calibration

    Yuhang Li, Shikuang Deng, Xin Dong, Ruihao Gong, and Shi Gu. A free lunch from ann: Towards efficient, accurate spiking neural networks calibration. InInternational confer- ence on machine learning, pages 6316–6325. PMLR, 2021. 7, 8

    2021

  20. [28]

    Spike calibration: Fast and accurate conversion of spiking neural network for object detection and segmenta- tion.arXiv preprint arXiv:2207.02702, 2022

    Yang Li, Xiang He, Yiting Dong, Qingqun Kong, and Yi Zeng. Spike calibration: Fast and accurate conversion of spiking neural network for object detection and segmenta- tion.arXiv preprint arXiv:2207.02702, 2022. 8

    2022 arXiv

  21. [29]

    Microsoft coco: Common objects in context

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Doll´ar, and C Lawrence Zitnick. Microsoft coco: Common objects in context. In Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceeding...

    2014

  22. [30]

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

    Xinhao Luo, Man Yao, Yuhong Chou, Bo Xu, and Guoqi Li. Integer-valued training and spike-driven inference spiking neural network for high-performance and energy-efficient object detection. InEuropean Conference on Computer Vi- sion, pages 253–272. Springer, 2024. 1, 2, 4, 6, 8

    2024

  23. [31]

    Training high- performance low-latency spiking neural networks by dif- ferentiation on spike representation

    Qingyan Meng, Mingqing Xiao, Shen Yan, Yisen Wang, Zhouchen Lin, and Zhi-Quan Luo. Training high- performance low-latency spiking neural networks by dif- ferentiation on spike representation. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, ...

    2022

  24. [32]

    Towards memory-and time-efficient backpropagation for training spiking neural networks

    Qingyan Meng, Mingqing Xiao, Shen Yan, Yisen Wang, Zhouchen Lin, and Zhi-Quan Luo. Towards memory-and time-efficient backpropagation for training spiking neural networks. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 6166–6176, 2023. 2, 7

    2023

  25. [33]

    Sign gradient descent-based neuronal dynamics: ANN-to-SNN conversion beyond reLU network

    Hyunseok Oh and Youngki Lee. Sign gradient descent-based neuronal dynamics: ANN-to-SNN conversion beyond reLU network. InForty-first International Conference on Machine Learning, 2024. 8

    2024

  26. [34]

    Gated attention coding for training high-performance and efficient spiking neural net- works

    Xuerui Qiu, Rui-Jie Zhu, Yuhong Chou, Zhaorui Wang, Liang-jian Deng, and Guoqi Li. Gated attention coding for training high-performance and efficient spiking neural net- works. InProceedings of the AAAI Conference on Artificial Intelligence, pages 601–610, 2024. 7, 8

    2024

  27. [35]

    Enabling deep spiking neural net- works with hybrid conversion and spike timing dependent backpropagation

    Nitin Rathi, Gopalakrishnan Srinivasan, Priyadarshini Panda, and Kaushik Roy. Enabling deep spiking neural net- works with hybrid conversion and spike timing dependent backpropagation. InInternational Conference on Learning Representations, 2020. 5, 7

    2020

  28. [36]

    Faster r-cnn: Towards real-time object detection with region proposal networks.IEEE transactions on pattern analysis and machine intelligence, 39(6):1137–1149, 2016

    Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun. Faster r-cnn: Towards real-time object detection with region proposal networks.IEEE transactions on pattern analysis and machine intelligence, 39(6):1137–1149, 2016. 1

    2016

  29. [37]

    Theory and tools for the conversion of ana- log to spiking convolutional neural networks.arXiv preprint arXiv:1612.04052, 2016

    Bodo Rueckauer, Iulia-Alexandra Lungu, Yuhuang Hu, and Michael Pfeiffer. Theory and tools for the conversion of ana- log to spiking convolutional neural networks.arXiv preprint arXiv:1612.04052, 2016. 2

    2016 arXiv

  30. [38]

    Going deeper in spiking neural networks: Vgg and residual architectures.Frontiers in neuroscience, 13:95,

    Abhronil Sengupta, Yuting Ye, Robert Wang, Chiao Liu, and Kaushik Roy. Going deeper in spiking neural networks: Vgg and residual architectures.Frontiers in neuroscience, 13:95,

  31. [39]

    Large language models encode clinical knowledge.Nature, 620 (7972):172–180, 2023

    Karan Singhal, Shekoofeh Azizi, Tao Tu, S Sara Mahdavi, Jason Wei, Hyung Won Chung, Nathan Scales, Ajay Tan- wani, Heather Cole-Lewis, Stephen Pfohl, et al. Large language models encode clinical knowledge.Nature, 620 (7972):172–180, 2023. 1

    2023

  32. [40]

    Deep directly-trained spik- ing neural networks for object detection

    Qiaoyi Su, Yuhong Chou, Yifan Hu, Jianing Li, Shijie Mei, Ziyang Zhang, and Guoqi Li. Deep directly-trained spik- ing neural networks for object detection. InProceedings of the IEEE/CVF International Conference on Computer Vi- sion, pages 6555–6565, 2023. 8

    2023

  33. [41]

    YOLOv5: A state-of-the-art real-time object de- tection system.https://docs.ultralytics.com,

    Ultralytics. YOLOv5: A state-of-the-art real-time object de- tection system.https://docs.ultralytics.com,

  34. [42]

    Attention is all you need.Advances in neural information processing systems, 30, 2017

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszko- reit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need.Advances in neural information processing systems, 30, 2017. 1

    2017

  35. [43]

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

    Yujie Wu, Lei Deng, Guoqi Li, Jun Zhu, Yuan Xie, and Lup- ing Shi. Direct training for spiking neural networks: Faster, larger, better. InProceedings of the AAAI conference on ar- tificial intelligence, pages 1311–1318, 2019. 2

    2019

  36. [44]

    Bkdsnn: Enhancing the performance of learning-based spiking neural networks training with blurred knowledge dis- tillation

    Zekai Xu, Kang You, Qinghai Guo, Xiang Wang, and Zhezhi He. Bkdsnn: Enhancing the performance of learning-based spiking neural networks training with blurred knowledge dis- tillation. InEuropean Conference on Computer Vision, pages 106–123. Springer, 2024. 7, 8

    2024

  37. [45]

    Spike-driven 10 transformer v2: Meta spiking neural network architecture in- spiring the design of next-generation neuromorphic chips

    Man Yao, Jiakui Hu, Tianxiang Hu, Yifan Xu, Zhaokun Zhou, Yonghong Tian, Bo Xu, and Guoqi Li. Spike-driven 10 transformer v2: Meta spiking neural network architecture in- spiring the design of next-generation neuromorphic chips. InInternational Conference on Learning Representations,

  38. [46]

    Scaling spike-driven transformer with efficient spike firing approximation training.IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025

    Man Yao, Xuerui Qiu, Tianxiang Hu, Jiakui Hu, Yuhong Chou, Keyu Tian, Jianxing Liao, Luziwei Leng, Bo Xu, and Guoqi Li. Scaling spike-driven transformer with efficient spike firing approximation training.IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025. 1, 2

    2025

  39. [47]

    Enhancing repre- sentation of spiking neural networks via similarity-sensitive contrastive learning

    Yuhan Zhang, Xiaode Liu, Yuanpei Chen, Weihang Peng, Yufei Guo, Xuhui Huang, and Zhe Ma. Enhancing repre- sentation of spiking neural networks via similarity-sensitive contrastive learning. InProceedings of the AAAI Conference on Artificial Intelligence, pages 16926–16934, 2024. 7, 8

    2024

  40. [48]

    Going deeper with directly-trained larger spiking neural net- works

    Hanle Zheng, Yujie Wu, Lei Deng, Yifan Hu, and Guoqi Li. Going deeper with directly-trained larger spiking neural net- works. InProceedings of the AAAI conference on artificial intelligence, pages 11062–11070, 2021. 7 11

    2021

  41. [2021]

    Accessed: insert date here. 8

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.