Author(s): Gilberto Reynoso-Meza; Elizabeth Pauline Carreno-Alvarado

Linked Author(s):

Keywords: Machine learning; Naive Bayes; Multi-objective optimisation; Water distribution systems

Abstract: Providing clean and safe drinking water is crucial for any water supply company. In such an activity, automatic anomaly detection plays a critical role in drinking water quality monitoring. In this sense, machine learning techniques have shown to be a powerful tool to extract or infer complex patterns from data. In the case of the so-called supervised learning, a given learner representation could learn such patterns using labeled data. For example, a helpful approach is to adjust a learner to detect anomalies. In this example, the learner's objective is to act as a binary classifier, where a balance between false negatives (predict a regular operation, when in fact an anomaly exists) and false positives (predict an anomaly, when there is not). A given learner attains such a balance via an optimization (learning phase), where its representation is adjusted. Multi-objective optimization techniques have a natural way of dealing with such problems because they perform a simultaneous optimization of conflicting objectives. This idea could be used in the training process of binary classifiers. This work analyzes a proposal where a multi-objective problem statement is stated for the Naive Bayes Classifier. Results in a competition benchmark for soft sensor proposals in water distribution systems show the usefulness of such an approach.

DOI: https://doi.org/10.3850/IAHR-39WC2521711920221690

Year: 2022