Seeing the Unseen: A Hybrid Supervised- Unsupervised Learning Framework for Zero-Day Cyber Threat Detection in Digital Transactions
Abstract: The rapid growth of digital transactions has significantly increased exposure to cyber threats, particularly zero-day attacks that do not follow previously known patterns. Conventional security systems largely depend on signature-based or fully supervised learning techniques, making them ineffective against evolving and unseen attack behaviors. This paper proposes a hybrid machine learning framework that integrates unsupervised anomaly detection with supervised classification to enhance the detection of both known and unknown cyber threats in digital transaction environments. The unsupervised component identifies abnormal transaction patterns without relying on labeled data, while the supervised component classifies known malicious activities using historical attack information. By combining both learning paradigms, the proposed framework improves adaptability, reduces dependency on labeled datasets, and enhances early threat detection capabilities. The performance of the approach is conceptually evaluated using standard security metrics such as accuracy, precision, recall, and false alarm rate. This work presents a flexible and scalable direction for strengthening digital transaction security and offers a strong foundation for future implementation and real-world validation.
Keywords: Zero-day attacks, Digital transactions, Anomaly detection, Hybrid learning, Machine learning security
How to Cite?: Noya Shaikh, Priya Budlani, Sonali Ingle, Vrishali Wable, "Seeing the Unseen: A Hybrid Supervised- Unsupervised Learning Framework for Zero-Day Cyber Threat Detection in Digital Transactions", Volume 15 Issue 4, April 2026, International Journal of Science and Research (IJSR), Pages: 94-96, https://www.ijsr.net/getabstract.php?paperid=SC26211105714, DOI: https://dx.doi.org/10.21275/SC26211105714