PROJECT TITLE :

Using Cost-Sensitive Learning and Feature Selection Algorithms to Improve the Performance of Imbalanced Classification

ABSTRACT:

The problem of unbalanced data is common in network intrusion detection, spam filtering, biomedical engineering, finance, and research, and it poses a barrier in a variety of real-world data-intensive applications. When typical classification methods are used to deal with unbalanced data, classifier bias occurs. The General Vector Machine (GVM) algorithm, as previously stated, has strong generalization capabilities, however it does not function well for imbalanced classification. Furthermore, the state-of-the-art Binary Ant Lion Optimizer (BALO) algorithm is highly exploitable and has a high rate of convergence. Based on these findings, we present a Cost-sensitive Feature Selection General Vector Machine (CFGVM) technique based on GVM and BALO algorithms to address the imbalanced classification problem by assigning different cost weights to distinct classes of data in this study. To increase classification performance, the BALO algorithm generates cost weights and extracts more significant features in our method. The CFGVM technique greatly enhances the classification performance of minority class samples, according to experiments done on eleven imbalanced data sets. When compared to other algorithms, including state-of-the-art methods, the suggested algorithm beats them in terms of performance and yields better classification results.