Peer-reviewed and preprint work on alignment, robustness, active learning, and industrial vision — six papers, four first-author (incl. one co-first), spanning ICML, AAAI, and arXiv.
Replaces DPO's frozen reference with a dynamic adversarial anchor — a worst-case local perturbation of the current policy — and reweights each preference pair by its geometric brittleness. Robust to noisy labels without an explicit noise model.
Diffusion synthesis that disentangles defect from background via masked cross-attention — the method behind the production system on the projects page.
Read the paper →Bayesian, data-dependent priors that let vision-language prompts adapt per input instead of staying fixed — better calibration under distribution shift.
Read the paper →An acquisition function that bridges sharpness-aware minimization and active learning — selecting samples that flatten the loss landscape.
Read the paper →Robustness to spurious group correlations via ordered risk and confidence regularization. Workshop on Spurious Correlations, Invariance and Stability (ICML 2022).
Read the paper →Deep variational clustering (VaDE) over sequences for credit-card delinquency prediction. IEEE Intl. Conference on Systems, Man, and Cybernetics.
Read the paper →