collaborators

10 papers

cs.CR2026

Model Confidence Under Answer-Preserving Attacks: An Informativeness-Manipulability Frontier

Reza Khanmohammadi, Ivan Brugere, Simerjot Kaur +3

Deployed vision-language systems often gate their answers on confidence, making confidence robustness relevant to oversight. We study confidence readouts under white-box, image-onl…

cs.CL2026

Confidence Estimation for Financial Vision-Language Models in Chart and Document Understanding

Reza Khanmohammadi, Simerjot Kaur, Charese H. Smiley +2

LVLMs are increasingly used to read financial charts, tables, and documents, where a single misread figure can move a decision and the most authoritative-looking answer is sometime…

cs.LG2026

Revisiting ML Training under Fully Homomorphic Encryption: Convergence Guarantees, Differential Privacy, and Efficient Algorithms

Yvonne Zhou, Mingyu Liang, Ivan Brugere +5

We present the first theoretical convergence analysis of machine learning training under fully homomorphic encryption (FHE), combined with a differentially private (DP) training al…

cs.CL2026

Grounded or Guessing? LVLM Confidence Estimation via Blind-Image Contrastive Ranking

Reza Khanmohammadi, Erfan Miahi, Simerjot Kaur +4

Large vision-language models suffer from visual ungroundedness: they can produce a fluent, confident, and even correct response driven entirely by language priors, with the image c…

cs.MA2026

MAFE: Enabling Equitable Algorithm Design in Multi-Agent Multi-Stage Decision-Making Systems

Zachary McBride Lazri, Anirudh Nakra, Ivan Brugere +5

Algorithmic fairness is often studied in static or single-agent settings, yet many real-world decision-making systems involve multiple interacting entities whose multi-stage action…

cs.LG2026

The Unseen Threat: Residual Knowledge in Machine Unlearning under Perturbed Samples

Hsiang Hsu, Pradeep Niroula, Zichang He +3

Machine unlearning offers a practical alternative to avoid full model re-training by approximately removing the influence of specific user data. While existing methods certify unle…