My publications are about de-anonymisation, identification and anonymisation, and in recent years increasingly about the privacy issues of face recognition. Most articles were published in English, the early ones in Hungarian.
The list is up to date, but the references and full bibliography are maintained elsewhere: Google Scholar (currently 542 citations, h-index 10), MTMT and ResearchGate. The code for the articles is on GitHub unless otherwise noted.
The Tenth International Conference on Cyber-Technologies and Cyber-Systems (CYBER 2025)
Advances on Societal Digital Transformation (DIGITAL 2025)
How to protect fingerprints derived from face recognition. The same question underlies the VeriDome product.
eTELEMED 2024
Development and evaluation of the Posture Buddy (in Hungarian: Tartás Pajtás) screening tool based on computer vision, within the IML4E project.
Egészségügyi Innovációs Szemle 3(3), pp. 40–55.
Testing the Posture Buddy tool supporting school nurse screening in a real environment, within the IML4E project.
Acta Cybernetica 25(2), pp. 233–255.
Comparison of freely downloadable face recognition libraries by privacy risk: what is revealed by tools anyone can use.
Automation and Applied Computer Science Workshop 2021 (AACS'21)
Budapest, Gondolat Kiadó, 2021. Chapter 2, pp. 19–171. (153 pages)
The publication includes the following three studies:
Study, commissioned by the Hungarian National Bank
The study, prepared for the Hungarian National Bank, examines how to estimate the risk of anonymising banking transaction data: whether the protection is strong enough, and how many data subjects could be affected by a successful re-identification.
Study, commissioned by the Hungarian National Bank
Infocommunications Journal 12(2), 50–56. o.
Shows that a face recognition imprint is not anonymous data: it can be reverse-engineered to reveal whose face it is.
12th Conference of PhD Students in Computer Science (CSCS 2020)
Automation and Applied Computer Science Workshop 2020 (AACS'20)
Workshop on Privacy in the Electronic Society (WPES'18)
Installed browser extensions and logged-in accounts together provide a unique fingerprint. The results were also presented by invitation at the European Parliament, and after publication, Google considered modifying Chrome.
Q1 International Journal of Information Security 18, pp. 125–139
The journal is currently ranked Q1; in the year of publication it was Q2. Post-peer-review, pre-copyedit version. The final, authenticated version is available from the publisher.
Workshop on Privacy in the Electronic Society (WPES'16), alongside ACM CCS'16
The Bumblebee attack: more correct re-identifications, fewer errors than previous methods. The attack can also be tried in the browser.
Proceedings on Privacy Enhancing Technologies, 2016/4
Transactions on Data Privacy 8(2), 113–140.
Doctoral dissertation, Budapest University of Technology and Economics
The doctoral dissertation. It is about how to defend against being identified solely based on your social network.
Successful defence: 13 May 2015. Supervisor: Sándor Imre, DSc, Budapest University of Technology and Economics, Department of Networked Systems and Services. Reviewers: Gergely Biczók and Julien Freudiger.
Privacy is a fundamental issue in social networks. The hardest problem is how much identifying information is carried by the metadata: the network's graph structure. Cleaned social network data is sometimes shared with third parties, business partners or researchers. Previous research has developed de-anonymisation attacks that re-identify users from this structure alone. The dissertation deals with these attacks and defences against them.
Periodica Polytechnica Electrical Engineering and Computer Science 58(4)
6th IEEE International Workshop on Security and Social Networking (SESOC), alongside PerCom
14th Joint IFIP TC6 and TC11 Conference on Communications and Multimedia Security (CMS 2013)
Research and Development in E-Business through Service-Oriented Solutions (eds. Tarnay K., Xu L., Imre S.), IGI Global
PinSoDa: Workshop on Privacy in Social Data, IEEE ICDM 2012, Brussels
The cited version is corrected: in Figure 3, the correlation is now calculated only for nodes that exist at the same time.
NordSec 2011, Tallinn, Estonia
My colleagues and I were the first to raise here that a user can be identified regardless of their browser. My most cited article.
Digital Identity and Access Management: Technologies and Frameworks, IGI Global
Infocommunications Journal III(4)
SECURWARE 2011, St-Laurent-du-Var, France
SECURWARE 2010, Venice, Italy
SECURWARE 2009, Athens, Greece
PET Convention 2009.1, Dresden, Germany
Joint SPACE and TIME Workshops, Trondheim, Norway, pp. 17–31.
Studies on Information and Knowledge Process 13. (ed. Ferenc Kiss)
Free Data, Protected Data 2. (ed. Iván Székely)
Free Data, Protected Data 2. (ed. Iván Székely)
Alma Mater series on information and knowledge processes 10. (ed. Ferenc Kiss), pp. 9–30.
Studies on Information and Knowledge Processes 11. (ed. Ferenc Kiss)