Utilising a Machine Learning Model with Feature Selection from the Salp Swarm Algorithm for Automated Spam Detection
P. Anandan, S. Manjula, M. Suganthy · 2023
Automated spam detection is a prevalent use case that involves the filtration and identification of unsolicited or undesirable messages, specifically spam comments or spam emails With the advent of mass mailing technology came an explosion of spam, and spam detection systems became an absolute must for keeping up with the problem. In order to identify spam using machine learning (ML), it is necessary to build a model that can separate incoming messages into two groups: spam and legitimate communications. This categorization relies on detecting specific traits and patterns derived from an appropriately annotated dataset. The dataset has provided these traits and trends. A number of approaches based on machine learning algorithms have been proposed for use in spam identification. Improving spam detection rates while lowering processing cost was the goal of developing algorithms for feature selection and parameter optimisation. The model described in this article is Auto-Spam Detection with Salp Swarm Algorithm based Feature Selection with Machine Learning (ASD-SSFSML). In order to detect and categorise spam, the ASD-SSFSML methodology employs feature selection and classification methods. The utilisation of technology allows for this to be achieved. This goal is ultimately achieved by employing the ASD-SSFSML technique. Improving Mini Batch K-Means Normalised Mutual Information enables feature extraction and optimum centroid selection with the use of the Fire Hawk Optimizer (FHO) algorithm. Applying the technique allows one to achieve this goal. In addition, the Salp Swarm (SS) method, when used for feature selection, improves classification accuracy while complicating training. More specifically, we use the Radial Bias Neural Network (RBNN) classifier to hunt for spammy emails. To ensure that the ASD-SSFSML method would yield superior outcomes, a comprehensive experimental validation process is executed. According to the results of the comparison, the ASD-SSFSML model is far more advanced than other, more recent models.