Pith. sign in

REVIEW 19 cited by

Optimal ANN-SNN Conversion for High-accuracy and Ultra-low-latency Spiking Neural Networks

Not yet reviewed by Pith; the record is open.

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

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2303.04347 v1 pith:7N3E5MIH submitted 2023-03-08 cs.NE

Tong Bu , Wei Fang , Jianhao Ding , PengLin Dai , Zhaofei Yu , Tiejun Huang This is my paper · ORCID
classification cs.NE
keywords snnsconversionactivationann-snnannsfunctiontime-stepsdatasets
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original abstract

Spiking Neural Networks (SNNs) have gained great attraction due to their distinctive properties of low power consumption and fast inference on neuromorphic hardware. As the most effective method to get deep SNNs, ANN-SNN conversion has achieved comparable performance as ANNs on large-scale datasets. Despite this, it requires long time-steps to match the firing rates of SNNs to the activation of ANNs. As a result, the converted SNN suffers severe performance degradation problems with short time-steps, which hamper the practical application of SNNs. In this paper, we theoretically analyze ANN-SNN conversion error and derive the estimated activation function of SNNs. Then we propose the quantization clip-floor-shift activation function to replace the ReLU activation function in source ANNs, which can better approximate the activation function of SNNs. We prove that the expected conversion error between SNNs and ANNs is zero, enabling us to achieve high-accuracy and ultra-low-latency SNNs. We evaluate our method on CIFAR-10/100 and ImageNet datasets, and show that it outperforms the state-of-the-art ANN-SNN and directly trained SNNs in both accuracy and time-steps. To the best of our knowledge, this is the first time to explore high-performance ANN-SNN conversion with ultra-low latency (4 time-steps). Code is available at https://github.com/putshua/SNN\_conversion\_QCFS

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 19 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Lapis: Laplacian Spiking Attention via First-Spike Timing and Membrane Leakage

    cs.NE 2026-08 conditional novelty 6.0 of 10

    Lapis replaces dot-product query-key scoring in spiking vision transformers with L1-distance-based Laplacian kernels on first-spike latencies, reaching near-dot-product accuracy at lower estimated arithmetic cost.

  2. AS-FedBridge: Pseudo-Spike Bridge Distillation for Heterogeneous ANN-SNN Federated Learning

    cs.LG 2026-08 conditional novelty 6.0 of 10

    AS-FedBridge trains a shared pseudo-spike bridge so ANN and SNN clients in federated learning can align representations and improve collaborative accuracy.

  3. Error Amplification Limits ANN-to-SNN Conversion in Continuous Control

    cs.NE 2026-01 conditional novelty 6.0 of 10

    Temporally correlated action errors, amplified by closed-loop dynamics, explain ANN-to-SNN conversion failures in continuous control, and cross-step residual potential initialization mitigates them.

  4. Ultra-Low-Latency Spiking Neural Networks with Temporal-Dependent Integrate-and-Fire Neuron Model for Objects Detection

    cs.CV 2025-08 reject novelty 6.0 of 10

    Binary temporal coding in a new spiking neuron (tdIF) plus delay-spike achieves strong ANN-to-SNN conversion results on detection and lane tasks, but the reported five-step latency ignores the depth-multiplied pipeline delay.

  5. FFGAF-SNN: The Forward-Forward Based Gradient Approximation Free Training Framework for Spiking Neural Networks

    cs.CV 2025-07 conditional novelty 6.0 of 10

    A Forward-Forward training framework that freezes spiking layers as black-box encoders and allocates channels by inter-class difficulty achieves 99.58% on MNIST, 92.13% on Fashion-MNIST, and 75.64% on CIFAR-10, the be...

  6. SPACT18: Spiking Human Action Recognition Benchmark Dataset with Complementary RGB and Thermal Modalities

    cs.CV 2025-07 conditional novelty 6.0 of 10

    SPACT18 is claimed to be the first action recognition dataset captured with a spike camera, paired with synchronized RGB and thermal video.

  7. LAS: Loss-less ANN-SNN Conversion for Fully Spike-Driven Large Language Models

    cs.LG 2025-05 conditional novelty 6.0 of 10

    LAS converts pre-trained ANN LLMs into fully spike-driven SNNs at 16 time steps using Outlier-Aware Threshold and Hierarchically Gated neurons, reporting near-lossless accuracy on OPT, BERT, GPT-2, and two vision-lang...

  8. Efficient Spiking Point Mamba for Point Cloud Analysis

    cs.CV 2025-04 conditional novelty 6.0 of 10

    Spiking Point Mamba combines Mamba state-space layers with spiking neurons and dynamic encoding to achieve state-of-the-art SNN accuracy on point cloud classification and part segmentation.

  9. FAS: Fast ANN-SNN Conversion for Spiking Large Language Models

    cs.LG 2025-02 conditional novelty 6.0 of 10

    FAS converts pretrained LLMs to spiking LLMs by fine-tuning with QCFS and then calibrating thresholds and initial membrane potentials, reaching near-LLM accuracy at 8-16 timesteps with large claimed energy savings.

  10. Self-Attentive Spatio-Temporal Calibration for Precise Intermediate Layer Matching in ANN-to-SNN Distillation

    cs.AI 2025-01 conditional novelty 6.0 of 10

    A self-attention mechanism aligns ANN and SNN feature layers across spatial and temporal dimensions, improving distilled SNN accuracy on static and neuromorphic benchmarks.

  11. Efficient and Robust Spiking Neural Networks for sEMG-Based Muscle Fatigue Detection

    cs.NE 2026-07 conditional novelty 5.0 of 10

    3- and 4-bit quantized SNNs with SDH training match ANN F1 on two sEMG fatigue datasets, stay more stable under seven noise types, and reduce estimated energy up to 201.77×.

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

    cs.NE 2026-03 conditional novelty 5.0 of 10

    Deep TTFS/latency-coded SNNs can be trained directly with backpropagation, reaching ~93.6% on CIFAR-10 with an average inference latency near one timestep.

  13. SDSNN: A Single-Timestep Spiking Neural Network with Self-Dropping Neuron and Bayesian Optimization

    cs.NE 2025-08 reject novelty 5.0 of 10

    A network using decay-triggered Self-Dropping neurons and Bayesian per-layer time-step search reports Fashion-MNIST 93.72%, CIFAR-10 92.20%, and CIFAR-100 69.45% accuracy with 21-56% energy savings versus multi-timestep LIF.

  14. STI-SNN: A 0.14 GOPS/W/PE Single-Timestep Inference FPGA-based SNN Accelerator with Algorithm and Hardware Co-Design

    cs.AR 2025-06 conditional novelty 5.0 of 10

    A single-timestep SNN inference accelerator on FPGA, co-designed with TET-based temporal pruning, reports competitive accuracy with lower latency and energy than two-timestep implementations.

  15. Efficient ANN-SNN Conversion with Error Compensation Learning

    cs.LG 2025-05 conditional novelty 5.0 of 10

    Using learnable clipped thresholds, half-threshold initialization, and negative-spike dual-threshold neurons, converted SNNs reach 94.75% on CIFAR-10 ResNet-18 at two time steps.

  16. Efficient Logit-based Knowledge Distillation of Deep Spiking Neural Networks for Full-Range Timestep Deployment

    cs.LG 2025-01 conditional novelty 5.0 of 10

    Temporal-wise logits distillation with ensemble self-distillation lets one SNN, trained at T=6, be deployed at T=1 through T=6 without retraining and with competitive accuracy.

  17. Self-cross Feature based Spiking Neural Networks for Efficient Few-shot Learning

    cs.LG 2025-05 conditional novelty 4.0 of 10

    A spike-based few-shot framework with self-correlation and cross-correlation modules achieves 98.9% on N-Omniglot 5w5s and competitive ANN-level accuracy on CUB and miniImageNet.

  18. Spiking Neural Network Feature Discrimination Boosts Modality Fusion

    cs.NE 2025-02 conditional novelty 4.0 of 10

    Applying L2 normalization to the final hidden features of audio and visual spiking networks, followed by spiking MLP fusion, yields 98.6% on CIFAR10-AV and 97.2% on UrbanSound8K-AV, outperforming a transformer-based S...

  19. Wafer2Spike: Spiking Neural Network for Wafer Map Pattern Classification

    cs.NE 2024-11 conditional novelty 4.0 of 10

    A spiking neural network achieves about 98% wafer-map defect classification accuracy on WM-811k, beating reported DNN baselines especially on rare defect patterns.

Pith tools