Towards Efficient Deployment of Hybrid SNNs on Neuromorphic and Edge AI Hardware

TL;DR

This paper presents a hybrid SNN-ANN system for dynamic vision data processing, deploying components on neuromorphic and edge hardware, and demonstrates superior performance over baseline models.

Contribution

The work introduces a hybrid SNN-ANN architecture, a novel data transfer circuit, and comprehensive hardware deployment and performance analysis on neuromorphic and edge AI devices.

Findings

Hybrid SNN-ANN outperforms baseline ANN in accuracy and latency.

Hybrid models consume less power and energy than baseline models.

Effective data transfer circuit enables seamless integration of spiking and non-spiking components.

Abstract

This paper explores the synergistic potential of neuromorphic and edge computing to create a versatile machine learning (ML) system tailored for processing data captured by dynamic vision sensors. We construct and train hybrid models, blending spiking neural networks (SNNs) and artificial neural networks (ANNs) using PyTorch and Lava frameworks. Our hybrid architecture integrates an SNN for temporal feature extraction and an ANN for classification. We delve into the challenges of deploying such hybrid structures on hardware. Specifically, we deploy individual components on Intel's Neuromorphic Processor Loihi (for SNN) and Jetson Nano (for ANN). We also propose an accumulator circuit to transfer data from the spiking to the non-spiking domain. Furthermore, we conduct comprehensive performance analyses of hybrid SNN-ANN models on a heterogeneous system of neuromorphic and edge AI hardware, evaluating accuracy, latency, power, and energy consumption. Our findings demonstrate that the hybrid spiking networks surpass the baseline ANN model across all metrics and outperform the baseline SNN model in accuracy and latency.