Optimization of AI Methods for Air Pollution Prediction

АУТОР(И) / AUTHOR(S): Goran KEKOVIĆ , Rade BOŽOVIĆ

Download Full Pdf   

DOI:  10.46793/ALFATECHproc25.078K

САЖЕТАК / ABSTRACT:

One of the biggest problems of large urban areas is air pollution, and in this regard, artificial intelligence (AI) methods can predict the level of pollution using a wide range of parameters. The use of artificial neural networks (ANN) based on Levenberg-Marquardt algorithm with Bayesian regularization (LMBR) is considered in this paper. It is shown that this algorithm achieves very high prediction accuracy, competitive with radial basis neural networks, which are commonly used for regression tasks. It is also shown that by choosing the optimal sample size, in addition to tuned ANN parameters, a balance can be achieved between the desired accuracy of the method and the deviation between simulated and real data. Relative error was used as a measure of that deviation. At the same time, it has been shown that sample size is not always a decisive factor affecting the efficiency of the AI method itself, but that a complete picture can be obtained by taking into account the entire structure of the input data.

КЉУЧНЕ РЕЧИ / KEYWORDS:

Air pollution; Artificial neural networks; Bayesian regularization; Levenberg – Marquardt algorithm

ПРОЈЕКАТ / ACKNOWLEDGEMENT:

ЛИТЕРАТУРА / REFERENCES:

• Chao, B. & Guang, Q.H. (2024). Air pollution concentration fuzzy evaluation based on evidence theory and the K-nearest neighbor algorithm. Frontiers of Environmental Science and Engineering 12, 147 https://doi.org/fenvs.2024.1243962.
• Liu, Y., Wen, L., Lin, Z. et al. (2024). Air quality historical correlation model based on time series. Scientific Reports 14, https://doi.org/10.1038/s41598-024-74246.
• Donnelly, A., Misstear, B., Broderick, B., (2015). Real time air quality forecasting using integrated parametric and non-parametric regression techniques. Atmos. Environ. 103, 53. https://doi.org/10.1016/j.atmosenv.2014.12.011.
• Biancofiore, F., Busilacchio, M., Verdecchia, M., Tomassetti, B., Aruffo, E., Bianco, S., Di Tommaso, S., Colangeli, C., Rosatelli, G., Di Carlo, P., (2017). Recursive neural network model for analysis and forecast of PM10 and PM2. 5. Atmospheric Pollution Research 8(4), 652-659. https://doi.org/10.1016/j.apr.2016.12.014.
• Huang, C.J., Kuo, P.H., (2018). A deep CNN-LSTM model for particulate matter (PM2.5) forecasting in smart cities. Sensors18 (7), 2220. https://doi.org/10.3390/s18072220.
• Cateni, S., Colla, V., Vannucci, M., (2023). Improving the Stability of the Variable Selection with Small Datasets in Classification and Regression Tasks. Neural Processing Letters55, 5331–5356. https://doi.org/10.1007/s11063-02210916-4.
• Guyon, I., Elisseeff, A., & Kaelbling, L. P. (Ed.). (2003). An introduction to variable and feature selection. Journal of Machine Learning Research, 3(7-8), 1157–1182. https://doi.org/10.1162/153244303322753616.
• Radovic, Ghalwash, M., Filipovic, N., Z. (2017). Minimum redundancy maximum relevance feature selection approach for temporal gene expression data. BMC Bioinformatics18(9), 2-14. https://doi.10.1186/s12859-016-1423-9.  
• Kohavi, R. & John, G.H. (1997). Wrappers for feature selection. Artif Intell 97(1–2), 273–324. https://doi.org/10.1016/s00043702(97)00043-x.
• Rodriguez-Galiano, V.F., Luque-Espinar, , Chica-Olmo, J.A.M., Mendes. M.P. (2018). Feature selection approaches for predictive modelling of groundwater nitrate pollution: An evaluation of filters, embedded and wrapper methods. Science of The Total Environment 624, 661-672. https://doi.org/10.1016/j.scitotenv.2017.12.152.  
• Singh, S., Giri, M., (2014). Comparative Study Id3, Cart And C4.5 Decision Tree Algorithm: A Survey. International Journal of Advanced Information Science and Technology (IJAIST) 3(7), https://doi:10.15693/ijaist/2014.v3i7.47-52.
• Uwimana, E., Zhou, Y. & Sall, N.M. A short-term load demand forecasting: Levenberg–Marquardt (LM), Bayesian regularization (BR), and scaled conjugate gradient (SCG) optimization algorithm analysis. J Supercomput  81, 55 (2025). https://doi.org/10.1007/s11227-02406513-y.
• Vito, S. (2008). Air Quality [Dataset]. UCI Machine Learning Repository. https://doi.org/10.24432/C59K5F.
• Vabalas, A, et al. (2019). Machine learning algorithm validation with a limited sample size. PLoS ONE14(11). https://doi.org/10.1371/journal.pone.0224365.
• Yiran S., Ding L.Y., Yantao T., Yaowu S. (2015). Air:fuel ratio prediction and NMPC for SIengines with modified Volterra model and RBF network, Appl. Artif. Int. 45, 313-324. https://doi.org/10.1016/j.engappai.2015.07.008.
• Rajput, D., Wang, W.J., Chen, C.C. (2023). Evaluation of a decided sample size in machine learning applications. Bioinformatics 24, 1-17. https://doi.org/10.1186/s12859-023-05156-9.