A Fully Spiking Hybrid Neural Network for Energy-Efficient Object\n Detection

Abstract

This paper proposes a Fully Spiking Hybrid Neural Network (FSHNN) for\nenergy-efficient and robust object detection in resource-constrained platforms.\nThe network architecture is based on Convolutional SNN using\nleaky-integrate-fire neuron models. The model combines unsupervised Spike\nTime-Dependent Plasticity (STDP) learning with back-propagation (STBP) learning\nmethods and also uses Monte Carlo Dropout to get an estimate of the uncertainty\nerror. FSHNN provides better accuracy compared to DNN based object detectors\nwhile being 150X energy-efficient. It also outperforms these object detectors,\nwhen subjected to noisy input data and less labeled training data with a lower\nuncertainty error.\n