Adaptive boundary-aware artificial immune recognition system for data classification
Year: 2024
Type: article
Abstract: In this paper, we present AIBARS (Adaptive Immune Boundary-Aware Recognition System), a novel learning algorithm that we have devised as an advancement over the traditional Artificial Immune Recognition System (AIRS). AIBARS is designed to address the key limitations of AIRS by integrating a boundary-aware stimulation measure and an antigen-specific stopping criterion, coupled with a more directed mutation approach. This innovation allows AIBARS to adeptly navigate complex decision boundaries between classes and substantially reduce the corresponding time complexity, especially in the context of high-dimensional datasets. Moreover, we incorporate a more targeted strategy for the mutation process that defies its intrinsically stochastic nature. The efficacy of AIBARS is rigorously evaluated in terms of classification accuracy, data reduction efficiency, and overall algorithmic performance through an extensive comparative analysis with the original AIRS classifier. For this evaluation, two datasets were employed: (1) synthetic data simulating a $C$-class classification problem in $L$-dimensional space and (2) 10.000 music pieces coming from 10 classes of Western music. The findings from this evaluation clearly indicate that our proposed method yields significant performance and efficiency gains over the original AIRS algorithm, while markedly reducing computational expenses.
Source: Information Sciences
Institution: University of Piraeus
Language: English
Subfield: Biomedical Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $3,360
OpenAlex ID: W4402859385