ACCELERATING SVM HYPERPARAMETER TUNING FOR PHISHING WEBSITE DETECTION USING HIGH-PERFORMANCE COMPUTING

AUTHOR(S)
Biljana Lakić^1*, Kristijan Kuk²

ABSTRACT
Phishing attacks continue to pose a significant threat to digital security, necessitating the development of more effective detection mechanisms. This study explores the optimization of widely used machine learning classifier, Support Vector Machines (SVM), for the task of phishing website detection, leveraging the computational capabilities of Serbia’s National AI Platform. The research focuses on hyperparameter tuning using optimization techniques executed in high-performance DGX-A100 nodes. By utilizing the FastML Engine within the Codex AI SUITE, the study achieves scalable orchestration of large-scale experiments, enabling rapid evaluation of numerous hyperparameter configurations. Results demonstrate substantial improvements in training time, highlighting the potential of HPC resources to enhance cybersecurity applications and support the strategic utilization of national AI infrastructure.

DOI

DOWNLOAD
https://unitech-selectedpapers.tugab.bg/images/2025/4-Computer%20system%20and%20technologies/p63_s4_u92_id477-SP.pdf

How to cite this article:
Biljana Lakić^1*, Kristijan Kuk², Yordan Petkov, ACCELERATING SVM HYPERPARAMETER TUNING FOR PHISHING WEBSITE DETECTION USING HIGH-PERFORMANCE COMPUTING, UNITECH – SELECTED PAPERS - 2025