Efthymios (Themis) Tsaprazlis

About Me

I'm a Computer Science PhD student at University of Southern California and a graduate research assistant at Signal Analysis and Interpretation Lab (SAIL), under the supervision of Professor Shrikanth Narayanan. I also collaborate closely with Professor Sai Praneeth Karimireddy. Previously, I did my Bachelors in the ECE Department at National Technical University of Athens where I worked with Professor Petros Maragos. Also, I was an external collaborator of the Machine Learning Lab at University of Texas at Austin, where I worked with Professor Alex Dimakis.

Research Interests

My research lies at the intersection of applied and theoretical machine learning, developing systems that learn efficiently, generalize across modalities, and remain trustworthy in real-world deployment.

News

• Jul 2026Our work "Can We Enable Speech Data With Differential Privacy?" will appear in Re-Data Workshop @ COLM 2026.
• Jun 2026Our paper "Rethinking Visual Privacy: A Compositional Privacy Risk Framework for Severity Assessment with VLMs" got accepted at ECCV 2026.
• Mar 2026Our paper "Assessing Visual Privacy Risks in Multimodal AI: A Novel Taxonomy-Grounded Evaluation of Vision-Language Models" got accepted at CVPR 2026 Workshop: SAIAD.
• Jan 2026Our paper "VoxGuard: Evaluating User and Attribute Privacy in Speech via Membership Inference Attacks" got accepted at ICASSP 2026.
• Nov 2025Received the Gerondelis Foundation Scholarship for my PhD studies at USC.
• Sep 2025Received a gift award from the USC-Amazon Center on Secure and Trusted Machine Learning for my research on explainable privacy understanding in multimodal AI.
• Jan 2025We got a paper accepted in ICRA 2025.
• Aug 2024I started my PhD at University of Southern California.
• Jan 2024I started a new position as Research Assistant in Robotics Unit of the Athena Research and Innovation Center (Athena RC) and CVSP Group.
• Oct 2023Our paper "Enhancing CLIP with a Third Modality" got accepted in NeurIPS 2023 Workshop: Self-Supervised Learning - Theory and Practice.
• Oct 2023I joined Workable as a Machine Learning Engineer intern.

Selected Publications

@inproceedings{tsaprazlis2026rethinking,
  title={Rethinking Visual Privacy: A Compositional Privacy Risk Framework for Severity Assessment with VLMs},
  author={Tsaprazlis, Efthymios and Feng, Tiantian and Ramakrishna, Anil and Karimireddy, Sai Praneeth and Gupta, Rahul and Narayanan, Shrikanth},
  booktitle={European Conference on Computer Vision},
  pages={269--286},
  year={2026},
  publisher={Springer Nature Switzerland}
}

@inproceedings{tsaprazlis2026voxguard,
  title={VoxGuard: Evaluating User and Attribute Privacy in Speech via Membership Inference Attacks},
  author={Tsaprazlis, Efthymios and Lertpetchpun, Thanathai and Feng, Tiantian and Karimireddy, Sai Praneeth and Narayanan, Shrikanth},
  booktitle={ICASSP 2026-2026 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},
  pages={19042--19046},
  year={2026},
  organization={IEEE}
}

@inproceedings{tsaprazlis2026assessing,
  title={Assessing Visual Privacy Risks in Multimodal AI: A Novel Taxonomy-Grounded Evaluation of Vision-Language Models},
  author={Tsaprazlis, Efthymios and Feng, Tiantian and Ramakrishna, Anil and Gupta, Rahul and Narayanan, Shrikanth},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={4292--4301},
  year={2026}
}

@inproceedings{tsaprazlis2023enhancing,
  title={Enhancing CLIP with a Third Modality},
  author={Tsaprazlis, Efthymios and Smyrnis, Georgios and Dimakis, Alexandros G and Maragos, Petros},
  booktitle={NeurIPS 2023 Workshop: Self-Supervised Learning - Theory and Practice},
  year={2023}
}

All publications

Contact

You can reach me via email at tsaprazl@usc.edu or connect with me on LinkedIn.