I’m Masoud
Poorghaffar Aghdam
I am a PhD candidate at Radboud University advised by Dr. Güneş Acar.
I am a researcher in the field of privacy.
Specifically, my work focuses on web privacy and privacy in ML.
In my free time, I enjoy exploring concepts and
ideas from other fields.
Education
$ ls ~/education --sort=recent
Ph.D. Candidate
Computer Science
Radboud University · Advisor: Dr. Güneş Acar
M.Sc.
Computer Engineering
Bilkent University · Advisor: Dr. Cicek
B.Sc.
Computer Engineering
University of Tabriz · Advisor: Dr. Tanha
Research Interests
$ cat ~/interests.txt
TBH, I like to explore everything related to computers!
Currently, I’m particularly interested in:
Digital Privacy
Tracking, malwaretising, and how scammer target search engines.
Data Privacy
Protecting private information across datasets and pipelines.
Privacy-Preserving ML
Training and deploying models without leaking the private data behind them.
Privacy in Healthcare
Proctecting private aspects of paitent data while maximizing the utility.
Academic Experience
$ history | grep academia
Bilkent University
Teaching Assistant — CS101 Algorithms and Programming I
Teaching Assistant — CS223 Digital Design
Community Service
IEEE Transactions on Computational Biology and Bioinformatics (TCBB) — Reviewer
Research in Computational Molecular Biology (RECOMB) — Speaker
Intelligent Systems for Molecular Biology (ISMB) Conference — Reviewer
Papers
Generated Data with Fake Privacy: Hidden Dangers of Fine-tuning Large Language Models on Generated Data
Authors: Akkus A, Poorghaffar Aghdam M, Li M, Chu J, Backes M, Zhang Y, Sav S.
Fine-tuning large language models (LLMs) with generated data is often considered a privacy-preserving alternative to real data, but our study reveals significant privacy risks. We evaluate Personal Information Identifier (PII) leakage and Membership Inference Attacks (MIAs) on the Pythia Model Suite and Open Pre-trained Transformer (OPT), finding that fine-tuning with generated data can increase privacy vulnerabilities.
USENIX Security '25 · https://usenix.org/conference/usenixsecurity25/presentation/akkus
2023/06/30
1