Publications
You can also find my publications on Google Scholar.
Visual Grounding and Deliberation Help Multi-Robot Teams Make High-Stakes Decisions
Mohammadreza Teymoorianfard, Saaduddin Mahmud, Jean-Philippe Monteuuis, Jonathan Petit, Shlomo Zilberstein, Amir Houmansadr, Eugene Bagdasarian
Under review 2026
A benchmark for multi-robot safety decisions under distributed information, and HUDDLE, a structured deliberation protocol that raises team accuracy to up to 96%.
Rethinking Image Watermarking in the Era of Frontier Generative Models
Ali Naseh, Mohammadreza Teymoorianfard, Amir Houmansadr
Under review 2026
An API-only regeneration attack with GPT-Image-2 that removes 89% of watermarks across 13 methods, a concept-extraction attack on IConMark, and K-SemMark-R, a registry-based content watermark that resists regeneration.
TRACE: Trajectory Reasoning through Adaptive Cross-Step Evidence Aggregation for LLM Agents
Vijitha Mittapalli, Shreyaa Jayant Dani, Satya Srujana Pilli, Snigdha Ansu, Mohammadreza Teymoorianfard, Franck Dernoncourt, Hongjie Chen, Yu Wang, Ryan A. Rossi, Nesreen K. Ahmed
arXiv preprint arXiv:2606.07054 2026
A monitoring framework for long-horizon LLM agent trajectories that links evidence across temporally distant actions through a Triage-Inspect-Judge loop.
ReasonBreak: Probing Vulnerabilities in Reasoning-Enabled Vision-Language-Action Models for Autonomous Driving
Mohammadreza Teymoorianfard, Jean-Philippe Monteuuis, Jonathan Petit, Amir Houmansadr
arXiv preprint arXiv:2605.29114 2026
The first systematic black-box study of reasoning-enabled VLA models for autonomous driving under realistic textual input corruptions, with a reasoning-aware evaluation framework and benchmark.
Network-Level Prompt and Trait Leakage in Local Research Agents
Hyejun Jeong, Mohammadreza Teymoorianfard, Abhinav Kumar, Amir Houmansadr, Eugene Bagdasarian
USENIX Security Symposium 2026
A prompt and user-trait leakage attack against locally deployed web research agents that uses only network-level metadata such as visited IP addresses and their timings.
VidStamp: A Temporally-Aware Watermark for Ownership and Integrity in Video Diffusion Models
Mohammadreza Teymoorianfard, Siddarth Sitaraman, Shiqing Ma, Amir Houmansadr
arXiv preprint arXiv:2505.01406 2025
A watermarking framework that embeds frame-level messages through the decoder of a latent video diffusion model, enabling high-capacity, imperceptible watermarks and precise temporal tamper localization.
MeanSparse: Post-Training Robustness Enhancement Through Mean-Centered Feature Sparsification
Sajjad Amini, Mohammadreza Teymoorianfard, Shiqing Ma, Amir Houmansadr
arXiv preprint arXiv:2406.05927 2024
A simple post-processing technique that sparsifies mean-centered features in adversarially trained models, setting new AutoAttack robustness records on CIFAR-10, CIFAR-100, and ImageNet.