A Survey on Ultra-Lightweight and Energy-Efficient Deep Learning Models for Resource-Constrained IoT Devices

Abstract

This survey of literature examines the recent developments in ultra-lightweight and energy-efficient deep learning models which are specifically created to fit devices with limited resources and that are deployed in the IoT. The survey examines the ways of reducing model size and optimization of computation and balancing security and energy efficiency. The research on lightweight encryption and object recognition proves the application of these models in low-power settings in practice. The questionnaire will serve to present an overall picture of the methods and algorithms that allow the effective application of deep learning to the limited Internet of Things devices.