Machine Learning and Deep Learning Algorithms for Smart Cities: A Start-of-the-Art Review

Abstract

The development in our urban cities has increased significant risks with everyday lives, like traffic congestion, pollution of the atmosphere, energy use, and public safety among others. Internet of Things (IoT) system has been used to tackle different research issues in a smart city. With the rapid development of IoT technologies, researchers have been motivated to develop smart services that extract knowledge from big data generated from IoT-based devices/sensors. The development of various models like forecast, preparation, monitoring, and ambiguity exploration in smart cities has been enhanced by the applications of deep learning (DL) and machine learning (ML) techniques, and for the urban development. These have also yielded greater results in the process of the huge data and input variables coming from IoT-based cognitive cities. Therefore, this chapter reviews the applicability of the state-of-the-art ML and DL in smart cities’ developments. It also discusses the novel application taxonomy of ML and DL smart cities and environmental planning that includes terms that are used interchangeably. Research shows that urban transportation, energy, and healthcare system are the main areas of applications that ML and DL techniques contributed in addressing their problems. The finding from the reviews reveals that ML and DL methods that are mostly applicable, and used in smart cities and urban development, are decision trees, support vector machine, artificial neural network, Bayesian, neuro-fuzzy, ensembles, and their hybridizations. Due to the complexities of both ML and DL with broad coverage of smart city applications, the study shows that there are various challenges ahead in applying these algorithms for this emerging field. The chapter discusses a range of potential directions related to ML and DL efficacy, evolving frameworks, convergence of information, and protection of privacy hoping that these would take the relevant research one step further to fully develop data analytics for smart cities.