Method of developing machine learning architecture for IoT­devices based on serverless architecture

DOI: 10.31673/2412-9070.2020.046880

Abstract

With the advent of IoT and microservice architectures, a multitude of intelligent distributed applications have emerged in which IoT devices collect, transform, and analyze data in large volumes and at high speed. A large number of these programs require robust, predictive analytics in real time, which require threading workflows closer to the data source, as well as dynamic resource management decisions. Moreover, predictive analytics requires the developer to create robust deep learning models, which, in turn, requires them to develop functions, find and configure parameters, and select machine learning models, which takes not only time, but also requires high experience level. The proliferation of machine learning libraries and frameworks, data ingestion tools, streaming and batch processing engines, rendering techniques, and the myriad of available hardware platforms further exacerbate these challenges. To overcome these complex challenges faced by developers of intelligent IoT applications, this article proposes a method for deploying machine learning architecture for IoT devices based on a serverless architecture called MLAbosa. The MLAbosa method can deploy, plan and dynamically manage data transfer tools, streaming software, batch analytics tools, and visualization tools across the cloud spectrum. This article describes the architecture of the MLAbosa method, highlighting the problems it solves.

Keywords: Internet of Things; quality of service; machine learning; artificial intelligence; deep neural networks.

References
1. Інтелектуальні машини на службі сучасного суспільства. URL: https://www.everest.ua/intelektualni-mashynyna-sluzhbi-suchasnogo-suspilstva-2/. (дата звернення 15.01.2020).
2. Deep learning applications and challenges in big data analytics / М. Najafabadi, F. Villanustre, M. Khosh [et al.] // Journal of Big Data. 2015. № 2(1):1. URL:
3. Бойко С. С чего начать работу с ML и DL. Обзор лучших библиотек. URL: https://dou.ua/lenta/articles/best-libraries-tostart-with-ml/. (дата звернення 23.02.2019).
4. Сервисы Amazon ML: что такое AWS Sage-Maker. URL: https://senior.ua/articles/servisy-amazon-mlchto-takoe-aws-sagemaker (дата звернення) 15.03.2020).
5. 10 прикладів, як штучний інтелект може змінити ваш спосіб життя. URL: https://www.radiosvoboda.org/a/29015231.html (дата звернення 23.05.2019).