REVIEW 5 major objections 6 minor 30 references
ILIF: Temporal Inhibitory Leaky Integrate-and-Fire Neuron for Overactivation in Spiking Neural Networks
T0 review · 5 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash
Pith's one-line read Two temporally coupled inhibitory units added to the leaky integrate-and-fire neuron are claimed to dissolve the surrogate-gradient trade-off between overactivation and vanishing gradients, cutting spike counts while raising accuracy on…
desk verdict Empirically solid neuron model with a genuinely new inhibitory architecture, but the theoretical proof that it resolves the gamma dilemma does not hold up as written. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The carrying object is the ILIF neuron: a vanilla leaky integrate-and-fire core wrapped in two recurrent inhibitory units. The membrane-potential inhibitory unit (MPIU) integrates post-spike potential into a slowly decaying state $\bar{U}^l[t] = \lambda_U\bigl(\bar{U}^l[t-1] + S^l[t]\cdot\bar{m}^l[t]\bigr)$ and subtracts a bounded sigmoid of that state, $S^l[t]\cdot\sigma\bigl(\bar{U}^l[t]\bigr)$, from the reset potential, mimicking afterhyperpolarization for long-term suppression. The current inhibitory unit (CIU), $\bar{I}^l[t] = \lambda_I\bigl(\bar{I}^l[t-1] + S^l[t]\cdot I^l[t]\bigr)$, decays quickly and subtracts the previous time step's current from the incoming synaptic current, mimicking fast retrograde feedback. These units do two jobs at once: forward, they suppress firing once activity accumulates; backward, their interconnections add the extra gradient terms $\varphi^l[t]$ to the backpropagation-through-time expansion and, per Theorem 3, provide a shortcut intended to carry temporal gradients without decay.
What would settle it
Compute the exact backward pass for a single ILIF neuron over $T$ time steps (or a tiny two-neuron network) and compare the true gradient $\partial L/\partial U^l[t]$ with the no-decay product asserted in Eq. (23); if the ratio of the two shrinks as $T$ grows, the shortcut does not exist. A cheaper experiment: train ILIF with $\gamma = V_{\mathrm{th}}$, the regime where Lemma 2 kills every temporal LIF gradient via $\epsilon = 1 - V_{\mathrm{th}}/\gamma = 0$; if ILIF accuracy also collapses there, its inhibitory pathways do not actually bypass the $\gamma$-limited product.
Extended reading notes
Core claim
The paper's central claim is that temporal inhibition, not surrogate tuning, resolves the $\gamma$ dilemma. With the two inhibitory units in place, a network can settle at equilibrium weights of larger magnitude than vanilla LIF (Theorem 1) while its equilibrium firing rate is strictly lower (Theorem 2), because inhibition grows with firing history and offsets the stronger excitatory drive. The paper further claims these units create direct backward shortcuts through time, so gradients travel from late to early time steps without the attenuating factor $1 - V_{\mathrm{th}}H'(U)$ that otherwise vanishes at small $\gamma$ (Theorem 3). Empirically, ILIF matches or exceeds prior neuron models on four datasets while firing less, including a more-than-30% reduction in continuous firing rate on DVSGesture, which the paper reads as evidence that the inhibition mechanism, not gradient tuning, carries the improvement.
Load-bearing premise
The theoretical resolution of the gradient-vanishing half of the $\gamma$ dilemma rests on Theorem 3's assertion that the backward chain through the inhibitory shortcut collapses to the single product $\frac{\partial U^l[T-1]}{\partial S^l[t]}\frac{\partial S^l[t]}{\partial U^l[t]}$ with no attenuation; that chain-rule cancellation is stated, not derived, and is not generally valid, so if it fails the claim that ILIF preserves gradient flow is no longer established by the theory, and the $\gamma$ dilemma is resolved only empirically.
Editorial extensions
If this is right
- If ILIF works as claimed, SNN training no longer needs to tune $\gamma$ against the firing-versus-vanishing trade-off: accuracy stays stable across $\gamma$ values while firing rates stay low.
- The accuracy gains (95.49% CIFAR10, 78.51% CIFAR100, 78.60% DVSCIFAR10, 97.92% DVSGesture) come with fewer spikes, which the paper's energy model translates into lower estimated consumption per forward pass on three of the four datasets.
- The inhibitory units are neuron-local and add no learnable parameters beyond the two decay constants, so they can be dropped into other spiking neurons; the paper's IPLIF variant improves on PLIF across all four datasets.
- Theorems 1 and 2 predict a decoupling of weight magnitude from firing rate, equilibrium weights grow while firing rates fall, which is a distinctive signature that other overactivation controls such as adaptive thresholds or normalization would not produce.
- Training stabilizes: loss curves are smoother and converge lower than LIF, consistent with the claimed extra gradient pathways and with the reduction in overactivation.
Reading between the lines
- Editorial extension: ILIF front-loads spiking activity (the first time step fires more, later steps less), so a testable consequence is that inference could be truncated after the first few time steps with little accuracy loss, enabling early-exit SNNs the paper does not discuss.
- Editorial extension: the inhibition acts as an activity-dependent self-regulator; freezing or ablating the MPIU saturation curve would test whether stability comes from the bounded sigmoid or from the gradient shortcut.
- Editorial extension: a LIF baseline with its threshold raised to match ILIF's firing rate would isolate the firing-rate effect from the gradient-pathway effect, separating regularization from the claimed shortcut mechanism.
- Editorial extension: because ILIF's benefit is claimed to grow as inhibition accumulates, the mechanism should matter most for long sequences and deep networks; measuring accuracy and firing-rate gaps versus sequence length is a direct stress test.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper identifies a trade-off in the surrogate gradient support width γ: large γ leads to overactivation and excessive firing, whereas small γ causes vanishing temporal gradients. The authors propose an ILIF neuron model with two interconnected inhibitory units, MPIU and CIU, and report experiments on CIFAR10, CIFAR100, DVSCIFAR10, and DVSGesture showing higher accuracy, lower firing rates, and in most cases lower energy consumption than LIF. They further claim a theoretical analysis (Lemmas 1–2 and Theorems 1–3) proving that ILIF overcomes the γ dilemma by reducing firing rates while preserving gradient flow.
Significance. The empirical part of the paper is valuable: the reported accuracies are competitive with recent SNN methods, the ablation in Table 2 supports the contribution of each inhibitory unit, Table 3 suggests the mechanism transfers to PLIF, and the code is made available. If the theoretical claims were adequately supported, the paper would make a clear conceptual contribution to SNN neuron design. However, the theoretical core as written does not establish the central claims: the proof of the no-decay gradient shortcut is invalid, Lemma 1 contains a variable slip and an unsupported weight-norm argument, Theorem 2 assumes the conclusion through the definition of its auxiliary functions, and the energy-efficiency claim is contradicted by the paper's own Table 5 on DVSGesture. The manuscript is therefore not acceptable in its current form, but the strengths of the empirical study justify a major revision rather than rejection.
major comments (5)
- [Section 4.3, Eq. (23)] The no-decay shortcut identity in Eq. (23) is not a valid chain-rule collapse for the ILIF dynamics. In Eq. (17), the membrane potential at a later time step depends on S_l[t] not only through U_l[t] but also through \bar m_l[t], the MPIU accumulator, and the CIU state, so ∂U_l[T−1]/∂S_l[t] is not equal to the telescoped product multiplied by ∂U_l[t]/∂S_l[t]; intermediate Jacobians do not generally cancel. Appendix C's expansion in Eq. (37) and Appendix D's Eq. (43) show that every additional term is still multiplied by products of λε_l[t], the same attenuation factors that vanish for small γ, so no attenuation-free pathway is exhibited. Consequently, Theorem 3's claim that ILIF preserves gradient flow through a direct shortcut is not established by the provided derivation.
- [Section 4.1, Lemma 1] The proof of Lemma 1 has a variable slip and a non-sequitur. It starts with γ_2 > γ_1 but then states that 'As λ increases from λ1 to λ2'; if λ is intended, it is never linked to the γ comparison. More importantly, the proof asserts without derivation that increasing the support width causes previously inactive neurons to re-enter the update interval and that this increases the equilibrium weight norm. The sign annotations in Eq. (12) do not supply the missing argument: the factor (Y_L[t′] − \hat Y[t′]) is not nonnegative in both cases, and no monotonicity of the equilibrium weight norm in γ is shown. Since Lemma 1 is used to define the core 'dilemma of γ', this needs to be repaired.
- [Section 4.3, Theorem 1 and Appendix D] Theorem 1 relies on two unproven sign properties: the statement that the additional gradient b(W) 'shares the same sign as W', and the claim in Appendix D, Eq. (43), that ∂m_l[t′]/∂U_l[t′] < 0. These properties are not derived from Eq. (17). Because the notation in Eq. (17) overloads U_l[t] for both the membrane potential and the MPIU accumulator, the sign of ∂m_l[t′]/∂U_l[t′] depends on which variable is meant and on the value of S_l[t′]; it is not negative in general. The assertion that a(W) 'typically' opposes the sign of W is also heuristic rather than a proof. Thus the inequality ∥W″∥ > ∥W′∥ is not established.
- [Section 4.3, Theorem 2] The proof of Theorem 2 assumes the conclusion through the construction of the auxiliary functions. The functions F, R, and J are defined with exactly the properties needed to force r″ < r′: R and J are strictly increasing, J(r′) = 0, and H(r) = R(F(∥W″∥) − J(r)) is therefore decreasing. These properties are stipulated rather than derived from the ILIF equations, so the theorem restates the intended outcome as an assumption. In particular, J(r′) = 0—that inhibition vanishes exactly at the LIF equilibrium firing rate—is not established by Eqs. (13)–(17). The theorem therefore does not provide an independent analysis of the ILIF dynamics.
- [Section 5.4, Table 5 and Abstract] The abstract claims that ILIF 'improves energy efficiency by reducing firing rates', but Table 5 reports that on DVSGesture ILIF consumes 1034.647 µJ versus 961.133 µJ for LIF, i.e., ILIF is less energy-efficient on that dataset. The text in Appendix H acknowledges this, but the abstract and conclusion do not. Since energy efficiency is presented as a principal contribution, the claim must either be restricted to the datasets and settings where it holds, or the paper must quantify when the doubled MAC cost of ILIF is not offset by reduced spike counts.
minor comments (6)
- [Equations (13) and (17)] The same symbols U_l[t] and I_l[t] are used for both the membrane potential and the input current and for the MPIU and CIU accumulators, respectively. This overloading makes the derivations in Section 4.3 and Appendix D very difficult to follow; please use distinct symbols for the inhibitory accumulators.
- [Section 4.1, Lemma 1 proof] Please replace 'λ increases from λ1 to λ2' with 'γ increases from γ1 to γ2', or define λ explicitly if it is meant to be a different quantity.
- [Eq. (12)] The underbrace annotations in Eq. (12) should be corrected: the factor (Y_L[t′] − \hat Y[t′]) is a signed difference, not a nonnegative quantity, and the sign analysis needs to be written out separately for the cases \hat Y[t′] = 1 and \hat Y[t′] = 0.
- [Algorithm 1 and Appendix E] The pseudocode is titled 'Main Fire Procedure for LIF Model', which appears to be a leftover from a LIF version; it should be titled for ILIF. The heading 'Pseudocode of ILIF mdoel' also contains a typo.
- [Section 5.1 and Table 4] The text states 'with λ = 1 and τ = 1.1', but τ does not appear elsewhere in the experimental setup and is not defined. Please either remove it or state what it denotes.
- [Tables 1–3 and Figures 3–5] Only single-run results are reported, with no standard deviations or number of seeds. Given that several gains over baselines are small (e.g., 95.49 vs. 95.41 on CIFAR10) and that DVSGesture ties at 97.92, adding repeated-run statistics or at least stating that results are single runs would strengthen the empirical claims.
Circularity Check
Theorem 3's no-attenuation proof assumes the shortcut it claims to establish; the core gamma-dilemma theory is partially circular, while the experiments stand independently.
-
self definitional
[Section 4.3, Theorem 3 proof, Eq. (23)]
"Because each inhibitory unit is interconnected and the decay coefficient is set to 1 in our experiments, there exists a shortcut in the backward pass given by ∂Ul[t]/∂Sl[t] ∂Sl[t]/∂Ul[t] ... Since there is no decay in these shortcuts, the gradient can flow from the time step T−1 back to any earlier step without attenuation."
The theorem's conclusion is that ILIF 'mitigates the vanishing gradient problem' via additional gradient pathways. The proof's load-bearing step is the assertion that a no-decay shortcut ∂Ul[t]/∂Sl[t]·∂Sl[t]/∂Ul[t] exists; Eq. (23) then collapses the entire temporal chain ∂Ul[T-1]/∂Ul[T-2]...∂Ul[t+1]/∂Ul[t]·∂Ul[t]/∂Sl[t]·∂Sl[t]/∂Ul[t] to ∂Ul[T-1]/∂Sl[t]·∂Sl[t]/∂Ul[t]. That collapse is not a consequence of the ILIF dynamics in Eq. (17): future membrane potentials depend on Sl[t] through \bar ml[t], Ul[t], and Il[t], so ∂Ul[T-1]/∂Sl[t] is not generally the telescoped product. The proof therefore assumes the no-attenuation pathway it claims to establish; the theoretical guarantee reduces to its own premise.
full rationale
The empirical part of the paper is self-contained: ILIF is evaluated against external benchmarks (CIFAR10/100, DVSCIFAR10, DVSGesture) and compared with published methods; no parameter is fitted to a subset and then 'predicted' on a closely related quantity. Lemmas 1 and 2 are informal trade-off arguments with no self-citation; Lemma 2 cites Huang et al. (external) only as background and derives the epsilon product itself. Theorems 1 and 2 are not data-fitting and, although their sign and fixed-point arguments are questionable (Appendix D's sign claims and the fixed-point footnote are not rigorously established), they do not reduce to their inputs by definition. However, Theorem 3's proof of gradient-flow preservation is circular: the no-decay shortcut is asserted as the premise, and Eq. (23) is the conclusion restated as a chain-rule collapse. Since this theorem is the theoretical basis for the abstract's claim that ILIF 'overcomes the gamma dilemma', the theoretical derivation is partially circular. The independent experimental results keep the paper from being entirely circular; the DVSGesture energy result in Table 5 contradicting the abstract's energy-efficiency claim is a correctness/factual issue, not circularity.
Assumptions & free parameters
free parameters (5)
- lambda_U (MPIU decay) =
1.0
- lambda_I (CIU decay) =
0.03 on CIFAR10/CIFAR100/DVSCIFAR10, 0.05 on DVSGesture
- gamma (surrogate gradient support width) =
Not explicitly fixed; standard default V_th with experiments varying gamma
- V_th (firing threshold) =
1 (implied)
- sigma (sigmoid nonlinearity) =
Bounded sigmoid
assumptions (6)
- standard math Backpropagation through time with the chain rule yields the spatial and temporal gradient decomposition in Eq. (7).
- domain assumption The surrogate gradient derivative H'(U) = (1/gamma)1(|U-V_th|<gamma/2) is an adequate stand-in for the spike derivative.
- ad hoc to paper At equilibrium the additional gradient b(W) shares the sign of W.
- ad hoc to paper In Theorem 2, F, R, and J are strictly increasing functions with J(r')=0 capturing ILIF dynamics.
- ad hoc to paper The chain-rule collapse in Eq. (23) holds, giving a no-decay backward shortcut.
- ad hoc to paper Larger gamma increases equilibrium weight norm, linking gamma to overactivation.
invented entities (2)
-
MPIU (membrane potential inhibitory unit)
-
CIU (current inhibitory unit)
Cite this review
Pith. "Pith review of ILIF: Temporal Inhibitory Leaky Integrate-and-Fire Neuron for Overactivation in Spiking Neural Networks." pith.science (2026). https://pith.science/paper/DVSQLX3R
@misc{pith2026250510371,
author = {Pith},
title = {Pith review of: ILIF: Temporal Inhibitory Leaky Integrate-and-Fire Neuron for Overactivation in Spiking Neural Networks},
year = {2026},
howpublished = {\url{https://pith.science/paper/DVSQLX3R}},
note = {Machine review of arXiv:2505.10371}
}
read the original abstract
The Spiking Neural Network (SNN) has drawn increasing attention for its energy-efficient, event-driven processing and biological plausibility. To train SNNs via backpropagation, surrogate gradients are used to approximate the non-differentiable spike function, but they only maintain nonzero derivatives within a narrow range of membrane potentials near the firing threshold, referred to as the surrogate gradient support width gamma. We identify a major challenge, termed the dilemma of gamma: a relatively large gamma leads to overactivation, characterized by excessive neuron firing, which in turn increases energy consumption, whereas a small gamma causes vanishing gradients and weakens temporal dependencies. To address this, we propose a temporal Inhibitory Leaky Integrate-and-Fire (ILIF) neuron model, inspired by biological inhibitory mechanisms. This model incorporates interconnected inhibitory units for membrane potential and current, effectively mitigating overactivation while preserving gradient propagation. Theoretical analysis demonstrates ILIF effectiveness in overcoming the gamma dilemma, and extensive experiments on multiple datasets show that ILIF improves energy efficiency by reducing firing rates, stabilizes training, and enhances accuracy. The code is available at github.com/kaisun1/ILIF.
Figures
Figures from the paper (4 more)
Reference graph
Works this paper leans on
-
[1]
Long short-term memory and learning-to-learn in net- works of spiking neurons.NeurIPS, 31,
[Bellecet al., 2018 ] Guillaume Bellec, Darjan Salaj, Anand Subramoney, Robert Legenstein, and Wolfgang Maass. Long short-term memory and learning-to-learn in net- works of spiking neurons.NeurIPS, 31,
work page 2018
-
[5]
[Denget al., 2023 ] Shikuang Deng, Hao Lin, Yuhang Li, and Shi Gu. Surrogate module learning: Reduce the gradi- ent error accumulation in training spiking neural networks. InICML, pages 7645–7657. PMLR,
work page 2023
-
[7]
Temporal effective batch normalization in spiking neural networks.NeurIPS, 35:34377–34390,
[Duanet al., 2022 ] Chaoteng Duan, Jianhao Ding, Shiyan Chen, Zhaofei Yu, and Tiejun Huang. Temporal effective batch normalization in spiking neural networks.NeurIPS, 35:34377–34390,
work page 2022
-
[8]
[Fanget al., 2020 ] Haowen Fang, Amar Shrestha, Ziyi Zhao, and Qinru Qiu. Exploiting neuron and synapse filter dy- namics in spatial temporal learning of deep spiking neural network.arXiv preprint arXiv:2003.02944,
arXiv 2020
-
[9]
Reducing information loss for spiking neural networks
[Guoet al., 2022 ] Yufei Guo, Yuanpei Chen, Liwen Zhang, YingLei Wang, Xiaode Liu, Xinyi Tong, Yuanyuan Ou, Xuhui Huang, and Zhe Ma. Reducing information loss for spiking neural networks. InECCV, pages 36–52. Springer,
work page 2022
-
[10]
Membrane potential batch normalization for spiking neural networks
[Guoet al., 2023 ] Yufei Guo, Yuhan Zhang, Yuanpei Chen, Weihang Peng, Xiaode Liu, Liwen Zhang, Xuhui Huang, and Zhe Ma. Membrane potential batch normalization for spiking neural networks. InICCV, pages 19420–19430,
work page 2023
-
[11]
Learning both weights and connections for efficient neural network.Advances in neural information processing systems, 28,
[Hanet al., 2015 ] Song Han, Jeff Pool, John Tran, and William Dally. Learning both weights and connections for efficient neural network.Advances in neural information processing systems, 28,
2015
-
[14]
[Liet al., 2021 ] Yuhang Li, Yufei Guo, Shanghang Zhang, Shikuang Deng, Yongqing Hai, and Shi Gu. Differen- tiable spike: Rethinking gradient-descent for training spik- ing neural networks.NeurIPS, 34:23426–23439,
work page 2021
Show all 30 references
-
[15]
Networks of spiking neu- rons: the third generation of neural network models.Neu- ral networks, 10(9):1659–1671,
[Maass, 1997] Wolfgang Maass. Networks of spiking neu- rons: the third generation of neural network models.Neu- ral networks, 10(9):1659–1671,
1997
-
[17]
Towards memory-and time-efficient backpropagation for training spiking neural networks
[Menget al., 2023 ] 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. InICCV, pages 6166– 6176,
2023
-
[18]
[Neftciet al., 2019 ] Emre O Neftci, Hesham Mostafa, and Friedemann Zenke. Surrogate gradient learning in spik- ing neural networks: Bringing the power of gradient-based optimization to spiking neural networks.IEEE Signal Pro- cessing Magazine, 36(6):51–63,
2019
-
[20]
Deep learning in spiking neural net- works.Neural networks, 111:47–63,
[Tavanaeiet al., 2019 ] Amirhossein Tavanaei, Masoud Gho- drati, Saeed Reza Kheradpisheh, Timoth ´ee Masquelier, and Anthony Maida. Deep learning in spiking neural net- works.Neural networks, 111:47–63,
2019
-
[21]
Autaptic synaptic circuit enhances spatio-temporal predictive learn- ing of spiking neural networks.PMLR,
[Wang and Yu, 2024] Lihao Wang and Zhaofei Yu. Autaptic synaptic circuit enhances spatio-temporal predictive learn- ing of spiking neural networks.PMLR,
2024
-
[22]
Temporal- coded spiking neural networks with dynamic firing thresh- old: Learning with event-driven backpropagation
[Weiet al., 2023 ] Wenjie Wei, Malu Zhang, Hong Qu, Am- mar Belatreche, Jian Zhang, and Hong Chen. Temporal- coded spiking neural networks with dynamic firing thresh- old: Learning with event-driven backpropagation. In ICCV, pages 10552–10562,
2023
-
[23]
Backpropagation through time: what it does and how to do it.Proceedings of the IEEE, 78(10):1550–1560,
[Werbos, 1990] Paul J Werbos. Backpropagation through time: what it does and how to do it.Proceedings of the IEEE, 78(10):1550–1560,
1990
-
[25]
Online train- ing through time for spiking neural networks.NeurIPS, 35:20717–20730,
[Xiaoet al., 2022 ] Mingqing Xiao, Qingyan Meng, Zong- peng Zhang, Di He, and Zhouchen Lin. Online train- ing through time for spiking neural networks.NeurIPS, 35:20717–20730,
2022
-
[26]
Glif: A unified gated leaky integrate-and-fire neuron for spiking neural networks.NeurIPS, 35:32160– 32171,
[Yaoet al., 2022 ] Xingting Yao, Fanrong Li, Zitao Mo, and Jian Cheng. Glif: A unified gated leaky integrate-and-fire neuron for spiking neural networks.NeurIPS, 35:32160– 32171,
2022
-
[27]
Superspike: Supervised learning in multi- layer spiking neural networks.Neural computation, 30(6):1514–1541,
[Zenke and Ganguli, 2018] Friedemann Zenke and Surya Ganguli. Superspike: Supervised learning in multi- layer spiking neural networks.Neural computation, 30(6):1514–1541,
2018
-
[28]
Fast and robust learning in spiking feed-forward neural networks based on intrinsic plasticity mechanism.Neurocomputing, 365:102–112,
[Zhanget al., 2019 ] Anguo Zhang, Hongjun Zhou, Xiumin Li, and Wei Zhu. Fast and robust learning in spiking feed-forward neural networks based on intrinsic plasticity mechanism.Neurocomputing, 365:102–112,
2019
-
[29]
Going deeper with directly-trained larger spiking neural networks
[Zhenget al., 2021 ] Hanle Zheng, Yujie Wu, Lei Deng, Yi- fan Hu, and Guoqi Li. Going deeper with directly-trained larger spiking neural networks. InAAAI, volume 35, pages 11062–11070,
2021
-
[30]
Spikformer: When spiking neural network meets transformer.arXiv preprint arXiv:2209.15425, 2022
[Zhouet al., 2022 ] Zhaokun Zhou, Yuesheng Zhu, Chao He, Yaowei Wang, Shuicheng Yan, Yonghong Tian, and Li Yuan. Spikformer: When spiking neural network meets transformer.arXiv preprint arXiv:2209.15425, 2022
2022 arXiv
-
[1990]
Direct training for spiking neu- ral networks: Faster, larger, better
[Wuet al., 2019 ] Yujie Wu, Lei Deng, Guoqi Li, Jun Zhu, Yuan Xie, and Luping Shi. Direct training for spiking neu- ral networks: Faster, larger, better. InAAAI, volume 33, pages 1311–1318,
2019
-
[1997]
Training high-performance low-latency spiking neural net- works by differentiation on spike representation
[Menget al., 2022 ] Qingyan Meng, Mingqing Xiao, Shen Yan, Yisen Wang, Zhouchen Lin, and Zhi-Quan Luo. Training high-performance low-latency spiking neural net- works by differentiation on spike representation. InCVPR, pages 12444–12453,
2022
-
[2015]
Clif: Complementary leaky integrate-and- fire neuron for spiking neural networks.ICML,
[Huanget al., 2024 ] Yulong Huang, Xiaopeng Lin, Hongwei Ren, Haotian Fu, Yue Zhou, Zunchang Liu, Biao Pan, and Bojun Cheng. Clif: Complementary leaky integrate-and- fire neuron for spiking neural networks.ICML,
2024
-
[2018]
Neuro- morphic brain-inspired computing with hybrid neural net- works
[Cai and Li, 2021] Zongyuan Cai and Xinze Li. Neuro- morphic brain-inspired computing with hybrid neural net- works. In2021 IEEE International Conference on Artifi- cial Intelligence and Industrial Design (AIID), pages 343–
2021
-
[2019]
Cirm-snn: Certainty interval reset mechanism spiking neuron for en- abling high accuracy spiking neural network.Neural Pro- cessing Letters, 55(6):7561–7582,
[Niu and Wei, 2023] Li-Ye Niu and Ying Wei. Cirm-snn: Certainty interval reset mechanism spiking neuron for en- abling high accuracy spiking neural network.Neural Pro- cessing Letters, 55(6):7561–7582,
2023
-
[2021]
Towards ultra low latency spiking neural networks for vision and sequential tasks using temporal pruning
[Chowdhuryet al., 2022 ] Sayeed Shafayet Chowdhury, Nitin Rathi, and Kaushik Roy. Towards ultra low latency spiking neural networks for vision and sequential tasks using temporal pruning. InECCV, pages 709–726. Springer,
2022
-
[2022]
In- vestigating current-based and gating approaches for ac- curate and energy-efficient spiking recurrent neural net- works
[Dampfhofferet al., 2022 ] Manon Dampfhoffer, Thomas Mesquida, Alexandre Valentian, and Lorena Anghel. In- vestigating current-based and gating approaches for ac- curate and energy-efficient spiking recurrent neural net- works. InInternational Conference on Artificial Neural N...
2022
-
[2023]
Bi- ologically inspired dynamic thresholds for spiking neural networks.NeurIPS, 35:6090–6103,
[Dinget al., 2022 ] Jianchuan Ding, Bo Dong, Felix Heide, Yufei Ding, Yunduo Zhou, Baocai Yin, and Xin Yang. Bi- ologically inspired dynamic thresholds for spiking neural networks.NeurIPS, 35:6090–6103,
2022
-
[2024]
Tab: Temporal ac- cumulated batch normalization in spiking neural networks
[Jianget al., 2024 ] Haiyan Jiang, Vincent Zoonekynd, Giu- lia De Masi, Bin Gu, and Huan Xiong. Tab: Temporal ac- cumulated batch normalization in spiking neural networks. InICLR,
2024
Reviewed August 15, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.