Title:   Fostering the use of Deep Learning on IoT Devices

Authors: Mrs. S.karpagavalli, Assistant Professor, Department of Computer Science, Government First Grade College, Hoskote-562114

Abstract: Deep learning can facilitate Internet of Things (IoT) devices to comprehend unstructured multimedia data and react intelligently to both user and environmental events, but it has stringent performance and power needs. The authors investigate two ways to successfully incorporate deep learning with low-power IoT products.The first way is to use lightweight deep learning models. Lightweight models are smaller and less complex than traditional deep learning models, which means they require less processing power and memory. The second way is to use edge computing. Edge computing is a distributed computing paradigm where computation is performed closer to the data source. This can reduce the amount of data that needs to be transferred to the cloud, which can also reduce power consumption. The authors evaluate both of these approaches on a real-world IoT application. They show that both approaches can be used to successfully incorporate deep learning with low-power IoT products.

Keywords: Deep learning, Environmental events, User events, Real-world application

DOI: 10.5281/zenodo.8136866

International Journal of Applied Pattern Recognition, 2023 Vol.7 No.1, pp.99 - 103

Received: 12 Nov 2021
Accepted: 18 Dec 2021
Published online: 10 Jan 2022