PASCAL: Precise and Efficient ANN- SNN Conversion using Spike Accumulation and Adaptive Layerwise Activation

2025-05-03

AI Summary

To address the high inference latency and accuracy degradation in ANN-to-SNN conversion, this paper proposes a mathematically equivalent low-latency conversion framework. First, it establishes rigorous equivalence between ANNs and SNNs under Quadratic Continuous-Function Spiking (QCFS) activation. Second, it introduces a layer-wise adaptive QCFS quantization strategy that dynamically allocates the minimal necessary number of time steps per layer. Third, it incorporates a spike accumulation mechanism to enhance temporal information utilization efficiency. Evaluated on ImageNet, the converted ResNet-34 SNN achieves ≈74% top-1 accuracy with only 1/64 the inference steps required by prior methods—significantly outperforming existing conversion approaches. The core contribution lies in the synergistic co-optimization of theoretically grounded equivalence modeling and hardware-friendly adaptive quantization, jointly achieving high accuracy and ultra-low latency.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Learning & Optimization for CV

Application Category

• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Spiking Neural Networks (SNNs) have been put forward as an energy-efficient alternative to Artificial Neural Networks (ANNs) since they perform sparse Accumulate operations instead of the power-hungry Multiply-and-Accumulate operations. ANN-SNN conversion is a widely used method to realize deep SNNs with accuracy comparable to that of ANNs.~citeauthor{bu2023optimal} recently proposed the Quantization-Clip-Floor-Shift (QCFS) activation as an alternative to ReLU to minimize the accuracy loss during ANN-SNN conversion. Nevertheless, SNN inferencing requires a large number of timesteps to match the accuracy of the source ANN for real-world datasets. In this work, we propose PASCAL, which performs ANN-SNN conversion in such a way that the resulting SNN is mathematically equivalent to an ANN with QCFS-activation, thereby yielding similar accuracy as the source ANN with minimal inference timesteps. In addition, we propose a systematic method to configure the quantization step of QCFS activation in a layerwise manner, which effectively determines the optimal number of timesteps per layer for the converted SNN. Our results show that the ResNet-34 SNN obtained using PASCAL achieves an accuracy of $approx$74% on ImageNet with a 64$ imes$ reduction in the number of inference timesteps compared to existing approaches.

Problem

• Minimizes accuracy loss in ANN-SNN conversion
• Reduces inference timesteps for efficient SNN operation
• Optimizes layerwise activation for better performance

Innovation

• Mathematically equivalent ANN-SNN conversion with QCFS
• Layerwise adaptive QCFS activation quantization
• Minimal inference timesteps for high accuracy