Abstract

An intrusion detection system is an innovative part of software that analyzes a network for instances of fraud or violations of web-based application policies. Usually, a security data and management mechanism were utilized for notifying any hazardous act or infraction. The goal of this research is to develop a novel machine learning (ML)-driven technique for log-based detection of intrusions in financial cloud-based apps. In the financial industry, we introduced a new Coupling Catboost with Chameleon Optimization Algorithm (CC-COA) for identifying web or cloud-based application assaults. To develop our suggested model, we collected a dataset that contains information from the websites of significant local economic banks that store log files of activity. The CatBoost method was used to manage categorical information. Whereas the COA is used to optimize hyperparameters. A range of indicators, including accuracy (95%), precision (94%), F1 Score (93.5%) and recall (93%) are used in the outcome assessment phase to evaluate the effectiveness of the proposed model. The experimental outcomes demonstrate that the mentioned model outperformed the others in identifying different kinds of assaults in cloud financial apps.