Publication Date: 2023/09/18
Abstract: This manuscript offers an extensive and methodical literature review concerning the application of machine learning (ML) methods in the burgeoning field of smart cities. The review spans across vital smart city domains such as energy management, healthcare, transportation optimization, security enhancement, and pollution control. In this research, we present a cutting- edge research methodology that introduces a state-of- the-art taxonomy, evaluation framework, and model performance analysis, categorizing ML algorithms into four principal classes: decision trees, support vector machines, artificial neural networks, and advanced machine learning techniques, encompassing hybrid models, ensembles, and Deep Learning. Our study reveals that hybrid models and ensembles consistently outperform other ML approaches, exhibiting a compelling combination of high accuracy and cost- effectiveness. In contrast, deep learning (DL) techniques showcase superior accuracy but demand substantial computational resources. Furthermore, all advanced ML methods exhibit relatively slower processing speeds compared to single methods. Notably, support vector machines (SVM) and decision trees (DT) consistently outperform artificial neural networks (ANN) across various metrics. However, the margin of superiority is negligible, suggesting that either SVM or DT may be employed effectively depending on specific application requirements.
Keywords: Smart cities, Machine Learning, Support vector machine, Decision Trees, Artificial Neural Network, urban sustainbility, Deep Learning, Single and Hybrid Models.
DOI: https://doi.org/10.5281/zenodo.8354765
PDF: https://ijirst.demo4.arinfotech.co/assets/upload/files/IJISRT23SEP247.pdf
REFERENCES