Ivans Lubenko
Themes:
Completed Projects:
Interests
I work in the area of machine learning based steganography. The focus is on maximising the output from the classifier because it remains a largely unexplored area. I aim to show that a wise choice of classifier combined with better use of steganalysis data can help with some typical steganalysis problems such as classification in mismatched data. I focus on empirical studies of online linear and near-linear classifiers in the context of detecting typical stego schemes in real-world data.
Selected Publications
-
Going from small to large data in steganalysis
Ivans Lubenko and Andrew D. Ker
In Media Watermarking‚ Security‚ and Forensics 2012. Vol. 8303 of Proc. SPIE. Pages 0M01−0M10. SPIE. 2012.
Details about Going from small to large data in steganalysis | BibTeX data for Going from small to large data in steganalysis | DOI (10.1117/12.910214) | Link to Going from small to large data in steganalysis
-
Steganalysis with mismatched covers: do simple classifiers help?
Ivans Lubenko and Andrew D. Ker
In Proceedings of the on Multimedia and security. Pages 11–18. New York‚ NY‚ USA. 2012. ACM.
Details about Steganalysis with mismatched covers: do simple classifiers help? | BibTeX data for Steganalysis with mismatched covers: do simple classifiers help? | DOI (10.1145/2361407.2361410) | Link to Steganalysis with mismatched covers: do simple classifiers help?
-
PROCEEDINGS OF THE OXFORD UNIVERSITY COMPUTING LABORATORY STUDENT CONFERENCE 2010
Programme Committee: Sara−Jane Dunn‚ Joe Loughry‚ Ivan Lubenko (chair) and Huy Vu
No. RR−10−22. OUCL. November, 2010.
Details about PROCEEDINGS OF THE OXFORD UNIVERSITY COMPUTING LABORATORY STUDENT CONFERENCE 2010 | BibTeX data for PROCEEDINGS OF THE OXFORD UNIVERSITY COMPUTING LABORATORY STUDENT CONFERENCE 2010 | Download (pdf) of PROCEEDINGS OF THE OXFORD UNIVERSITY COMPUTING LABORATORY STUDENT CONFERENCE 2010