Title: Explainable AI and sand cat optimisation algorithm for water quality classification

Authors: Gehad Ismail Sayed; Aboul Ella Hassanien

Addresses: School of Computer Science, Canadian International College (CIC), Cairo, Egypt; Scientific Research School of Egypt (SRSEG), Egypt ' College of Business Administration (CBA), Kuwait University, Kuwait; Faculty of Computers and AI, Cairo University, Giza, Egypt; Scientific Research School of Egypt (SRSEG), Egypt

Abstract: Assessing river water quality is considered a critical task in enhancing water resource management plans. Therefore, an accurate prediction of the quality of the water has become highly needed to control water pollution. In this paper, a new water quality classification model is proposed based on explainable artificial intelligence (XAI) and an optimised artificial neural network (ANN). The sand cat optimisation algorithm (SCOA) is modified and applied for hyper-parameter optimisation of ANN. The proposed model is tested on a benchmark dataset of water quality taken from various places across India. The results are explained and interpreted using the XAI technique. The experimental results demonstrated that the modified SCOA can effectively find the optimal values of weights and bias coefficients for ANN. The proposed model can effectively classify the water quality. It obtained an overall accuracy of 98%, specificity of 99%, precision of 98%, sensitivity of 98%, and f-score of 98%.

Keywords: sustainable development goals; SDGs; machine learning algorithms; water quality index; water quality classification; swarm intelligence; sand cat optimisation algorithm; SCOA; metaheuristic optimisation algorithms; explainable artificial intelligence; XAI; artificial neural network; ANN; hyperparameters optimisation.

DOI: 10.1504/IJIEI.2024.137710

International Journal of Intelligent Engineering Informatics, 2024 Vol.12 No.1, pp.60 - 84

Received: 16 Sep 2023
Accepted: 13 Jan 2024
Published online: 02 Apr 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article