A Free Lunch From ANN: Towards Efficient, Accurate Spiking Neural Networks Calibration

TL;DR

This paper introduces a simple yet highly effective calibration method for converting pre-trained ANNs to SNNs, significantly improving accuracy with minimal data and computational effort, enabling state-of-the-art results on large-scale datasets.

Contribution

The paper presents a novel calibration algorithm that corrects conversion errors layer-by-layer, enhancing SNN accuracy derived from pre-trained ANNs with minimal additional training.

Findings

Achieves up to 69% top-1 accuracy improvement on ImageNet with MobileNet.

Effective calibration with few training samples and minutes of computation.

Enables state-of-the-art SNN performance on large-scale architectures.

Abstract

Spiking Neural Network (SNN) has been recognized as one of the next generation of neural networks. Conventionally, SNN can be converted from a pre-trained ANN by only replacing the ReLU activation to spike activation while keeping the parameters intact. Perhaps surprisingly, in this work we show that a proper way to calibrate the parameters during the conversion of ANN to SNN can bring significant improvements. We introduce SNN Calibration, a cheap but extraordinarily effective method by leveraging the knowledge within a pre-trained Artificial Neural Network (ANN). Starting by analyzing the conversion error and its propagation through layers theoretically, we propose the calibration algorithm that can correct the error layer-by-layer. The calibration only takes a handful number of training data and several minutes to finish. Moreover, our calibration algorithm can produce SNN with state-of-the-art architecture on the large-scale ImageNet dataset, including MobileNet and RegNet. Extensive experiments demonstrate the effectiveness and efficiency of our algorithm. For example, our advanced pipeline can increase up to 69% top-1 accuracy when converting MobileNet on ImageNet compared to baselines. Codes are released at https://github.com/yhhhli/SNN_Calibration.