MACHINE LEARNING APPROACHES IN ODOUR CLASSIFICATION ALGORITHM: A SYSTEMATIC LITERATURE REVIEW

Abstract

Odour classification using electronic nose (e-nose) has gained increasing importance due to its wide range of applications in environmental monitoring, food safety, and medical diagnostics. Recent advances in sensor technology and machine learning have significantly improved odour profiling capabilities; however, variations in methodology and evaluation approaches across studies remain a challenge. This systematic literature review aims to evaluate and synthesize existing classification algorithms applied in odour analysis, with particular emphasis on recent trends, limitations, and research opportunities. Following PRISMA guidelines, 27 primary studies published between 2020 and early 2024 were selected from IEEE Xplore, Scopus, and ScienceDirect for detailed analysis. This review examines key aspects of odour classification research, including sensor types, performance metrics, application domains, and machine learning algorithms. The findings show that traditional machine learning models such as Support Vector Machines (SVM) and Random Forest continue to be widely adopted due to their robustness, interpretability, and reliable performance when dealing with high-dimensional sensor data. In comparison, deep learning approaches such as Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) models demonstrate promising performance in more complex and large-scale classification tasks, although their use is often limited by higher computational requirements and data dependency. Feature extraction and feature selection are identified as critical factors influencing classification accuracy across different studies. This review further provides a structured synthesis linking sensor configurations, algorithm selection, and application domains, highlighting why certain approaches outperform others under specific constraints. Based on these insights, the study emphasizes the need for closer interdisciplinary collaboration between sensor developers and algorithm designers to support the development of scalable and real-world-ready odour classification systems.