Abstract

Financial cloud services have many improvements, but there are also deliberate security dangers, one of which is spamming. Furthermore, changing the spamming approach may have an impact on the model's efficiency and necessitate growing upgrades and developments. In this research, we introduce a unique technique, Dynamic Transient Search-driven Random Forest (DTS-RF), to advance and execute a capable Spam Detection System (SDS) that leverages to augment the security of financial transactions within cloud platforms. Research collected the dataset as an extensive collection of both legitimate and spam-related activity. A crucial step in getting the dataset prepared for SDS training is data cleansing. It requires eliminating noise, inconsistencies and extraneous information from the data to ensure model predictions are specific and dependable. The most pertinent and instructive characteristics are identified for the SDS's training by applying Independent Component Analysis (ICA) to feature extraction. Based on the suggested methodology, this study is carried out using the Python program and performance is examined in terms of accuracy (94%), precision (95.7%), recall (92.7%) and F1-score (93.5%). The solution achieves strong performance and increases the total resilience of financial networks towards spamming threats by combining DTS-RF.