Creating Proactive Cyber Threat Intelligence with Hacker Exploit Labels: A Deep Transfer Learning Approach

Benjamin M. Ampel, Sagar Samtani, Hongyi Zhu, Hsinchun Chen
Last updated on Oct 24, 2026
PDF DOI
Research design diagram: data collected from hacker forums, exploit dark-net markets and public exploit repositories is pre-processed into a gold-standard dataset, labeled by the Deep Transfer Learning Exploit Labeler (convolutional, BiLSTM and self-attention layers transferred from a source to a target domain), then evaluated in four experiments with benchmark and attention comparisons. DTL-EL Research Design
In Plain Terms

This study introduces DTL-EL, a machine-learning system that automatically reads the exploits hackers post in online forums and labels what kind of threat each one is. To make its labels more accurate, it borrows knowledge from related sources such as darknet marketplaces and public exploit databases. In testing, it classified exploits more accurately than existing methods, helping defenders prioritize the most dangerous threats faster.

Watch a short video summary

Artifacts

PDF DOI

Related Papers

Cite this

APA Ampel, B. M., Samtani, S., Zhu, H., & Chen, H. (2024). Creating Proactive Cyber Threat Intelligence with Hacker Exploit Labels: A Deep Transfer Learning Approach. MIS Quarterly, 48(1), 137–166. https://doi.org/10.25300/MISQ/2023/17316
BibTeX
@article{ampel2024creating,
  title={Creating Proactive Cyber Threat Intelligence with Hacker Exploit Labels: A Deep Transfer Learning Approach},
  author={Benjamin M. Ampel and Sagar Samtani and Hongyi Zhu and Hsinchun Chen},
  journal={MIS Quarterly},
  year={2024},
  doi={10.25300/MISQ/2023/17316}
}
Cyber Threat Intelligence Computational Design Science Deep Learning Exploit Labeling
Benjamin M. Ampel

Benjamin M. Ampel

Assistant Professor in Computer Information Systems and Director, Center for CyberAI Research (CCAIR)

My research focuses on AI-enabled Cybersecurity, including Cyber Threat Intelligence, Large Language Models, and Phishing Detection.