Vasisht Duddu
Distinguished Postdoctoral Fellow
Vector Institute for Artificial Intelligence
vasisht.duddu at outlook.com
Home Publications Awards Talks Mentoring Service CV
For non-first author publications: → Project lead (proposed idea, mentorship, implementation support, writing)
Pre-Prints
- PAL*M: Property Attestation for Large Generative Models Paper Code
Prach Chantasantitam, Adam Caulfield, Vasisht Duddu, Lachlan J. Gunn, N. Asokan
Under Submission - Contextualizing Tradeoffs in Automated Decision-Making Systems Paper Code
Rabeya Bosri, Anna Lorimer, Afrida Hossain, Vasisht Duddu, Bailey Kacsmar
Under Submission
2026
- SoK: Colluding Adversaries in Machine Learning Pipelines Paper Code
Vasisht Duddu, Lipeng He, Asim Waheed, N. Asokan
USENIX Security Symposium (USENIX SEC) - Locket: Robust Feature-Locking Technique for Language Models Paper Code
Lipeng He, Vasisht Duddu, N. Asokan
Association for Computational Linguistics (ACL) - PATCH: Mitigating PII Leakage in Language Models with Privacy-Aware Circuit Patching Paper Code
Anthony Hughes, Vasisht Duddu, N. Asokan, Nikolaos Aletras, Ning Ma
Findings of the European Chapter of ACL (EACL Findings) - Privacy Bias in Language Models: A Contextual Integrity-based Auditing Metric Paper
Yan Shvartzshnaider, Vasisht Duddu
Privacy Enhancing Technologies Symposium (PETS)
AAAI Workshop on Privacy-Preserving Artificial Intelligence (AAAI-PPAI) Oral Presentation
2025
- Combining Machine Learning Defenses without Conflicts Paper Code
Vasisht Duddu, Rui Zhang, N. Asokan
Transactions on Machine Learning Research (TMLR) - Position: Contextual Integrity is Inadequately Applied to Language Models Paper
Yan Shvartzshnaider*, Vasisht Duddu* (equal contribution)
International Conference on Machine Learning (ICML)
- Laminator: Verifiable ML Property Cards using Hardware-assisted Attestations Paper Code Poster@IEEE S&P
Vasisht Duddu, Oskari Järvinen, Lachlan J. Gunn, N. Asokan
ACM Conference on Data and Application Security and Privacy (CODASPY) - Espresso: Robust Concept Filtering in Text-to-Image Models Paper Code
Anudeep Das, Vasisht Duddu, Rui Zhang, N. Asokan
ACM Conference on Data and Application Security and Privacy (CODASPY) Best Paper Award
2024
- SoK: Unintended Interactions among Machine Learning Defenses and Risks Paper Code Blog
Vasisht Duddu, Sebastian Szyller, N. Asokan
IEEE Symposium on Security and Privacy (S&P) Distinguished Paper Award - GrOVe: Ownership Verification of Graph Neural Networks using Embeddings Paper Code
Asim Waheed, Vasisht Duddu, N. Asokan
IEEE Symposium on Security and Privacy (S&P) - Attesting Distributional Properties of Training Data for Machine Learning Paper Code
Vasisht Duddu, Anudeep Das, Nora Khayata, Hossein Yalame, Thomas Schneider, N. Asokan
European Symposium on Research in Computer Security (ESORICS)
2023
- Comprehension from Chaos: What Users Understand and Expect from Private Computation Paper Code
Bailey Kacsmar, Vasisht Duddu, Kyle Tilbury, Blase Ur, Florian Kerschbaum
ACM Conference on Computer and Communications Security (CCS)
2022
- Inferring Sensitive Attributes from Model Explanations Paper Code
Vasisht Duddu, Antoine Boutet
ACM International Conference on Information and Knowledge Management (CIKM)
2021
- Good Artists Copy, Great Artists Steal: Model Extraction Attacks Against Image Translation GANs Paper
Sebastian Szyller, Vasisht Duddu, Tommi Gröndahl, N. Asokan
Technical Report
2020
- Quantifying Privacy Leakage in Graph Embedding Paper Code
Vasisht Duddu, Antoine Boutet, Virat Shejwalkar
EAI International Conference on Mobile and Ubiquitous Systems (MobiQuitous)
NeurIPS Workshop on PPML
Software
- Amulet: A Python Library for Assessing Interactions Among ML Defenses and Risks Report Code
Asim Waheed, Vasisht Duddu, Rui Zhang, Sebastian Szyller
Theses
- Deployment Concerns in Machine Learning Systems: Unintended Interactions and Accountability Ph.D. Thesis
Vasisht Duddu (University of Waterloo, 2026) - Towards Effective Measurement of Membership Privacy Risk for Machine Learning Models Master's Thesis
Vasisht Duddu (University of Waterloo, 2022)
Technical Report → SHAPr: An Efficient and Versatile Membership Privacy Risk Metric for Machine Learning