Adaptive Calibration: A Unified Conversion Framework of Spiking Neural Networks

TL;DR

A novel Adaptive-Firing Neuron Model (AdaFire) is proposed, which dynamically adjusts firing patterns across different layers to substantially reduce the Unevenness Error - the primary source of error of converted SNNs within limited inference timesteps.

Abstract

Spiking Neural Networks (SNNs) have emerged as a promising energy-efficient alternative to traditional Artificial Neural Networks (ANNs). Despite this, bridging the performance gap with ANNs in practical scenarios remains a significant challenge. This paper focuses on addressing the dual objectives of enhancing the performance and efficiency of SNNs through the established SNN Calibration conversion framework. Inspired by the biological nervous system, we propose a novel Adaptive-Firing Neuron Model (AdaFire) that dynamically adjusts firing patterns across different layers, substantially reducing conversion errors within limited timesteps. Moreover, to meet our efficiency objectives, we propose two novel strategies: an Sensitivity Spike Compression (SSC) technique and an Input-aware Adaptive Timesteps (IAT) technique. These techniques synergistically reduce both energy consumption and latency during the conversion process, thereby enhancing the overall efficiency of SNNs. Extensive experiments demonstrate our approach outperforms state-of-the-art SNNs methods, showcasing superior performance and efficiency in 2D, 3D, and event-driven classification, as well as object detection and segmentation tasks.

Figures (10)

• Figure 1: Performance comparison on different tasks. Our method significantly outperforms the traditional Calibration method across all evaluated tasks, effectively narrowing the performance gap with ANNs while requiring limited timesteps.
• Figure 2: The workflow of the Pareto Frontier-driven search algorithm for automatically searching the optimal configurations of each layer in SNNs.
• Figure 3: Sensitivity result of each layer in ResNet-18. Subfigure(a): Sensitivity when using different max firing time $\varphi$ for each layer. Subfigure(b): Sensitivity when using different threshold ratios $\rho$ for each layer.
• Figure 4: Pareto Frontier Representation. Each data point represents a distinct layer-specific configuration.
• Figure 5: Adaptive-Firing Mechanism. Adjusting the maximum firing times $\varphi$ minimizes residual information, thereby decreasing conversion errors.