Theory and Tools for the Conversion of Analog to Spiking Convolutional Neural Networks

TL;DR

This paper introduces a new theoretical framework and practical tools for converting deep CNNs into spiking neural networks, enabling efficient, near-lossless inference suitable for real-time applications.

Contribution

The authors develop a novel theory explaining successful CNN-to-SNN conversion and introduce new mechanisms and implementations for common CNN operations to improve conversion accuracy.

Findings

Achieved state-of-the-art SNN performance on MNIST and CIFAR10.

Identified and fixed key sources of approximation errors in conversion.

Developed nearly lossless conversion methods for complex CNN architectures.

Abstract

Deep convolutional neural networks (CNNs) have shown great potential for numerous real-world machine learning applications, but performing inference in large CNNs in real-time remains a challenge. We have previously demonstrated that traditional CNNs can be converted into deep spiking neural networks (SNNs), which exhibit similar accuracy while reducing both latency and computational load as a consequence of their data-driven, event-based style of computing. Here we provide a novel theory that explains why this conversion is successful, and derive from it several new tools to convert a larger and more powerful class of deep networks into SNNs. We identify the main sources of approximation errors in previous conversion methods, and propose simple mechanisms to fix these issues. Furthermore, we develop spiking implementations of common CNN operations such as max-pooling, softmax, and batch-normalization, which allow almost loss-less conversion of arbitrary CNN architectures into the spiking domain. Empirical evaluation of different network architectures on the MNIST and CIFAR10 benchmarks leads to the best SNN results reported to date.