Statistics

Evaluating large language models increasingly relies on LLM-as-a-judge protocols, but such evaluations remain costly: different judges …

We study a new risk-averse bandit setting motivated by semiconductor manufacturing, where the quality of a recipe is judged not by its …

We revisit regularized regret minimization under full-information and bandit feedback, where a learner optimizes an objective of the …

We present GL-LowPopArt, a novel Catoni-style estimator for generalized low-rank trace regression. Building on LowPopArt (Jang et al., …

We study the problem of clustering T trajectories of length H, each generated by one of K unknown ergodic Markov chains over a finite …

Active learning improves training efficiency by selectively querying the most informative samples for labeling. While it naturally fits …

We consider the problem of regularized best-response max-regret minimization in online RLHF under general preferences and bandit …

We consider the problem of heteroskedastic generalized linear bandits (GLBs) with adversarial corruptions, which subsumes various …

We present a unified likelihood ratio-based confidence sequence (CS) for any (self-concordant) generalized linear model (GLM) that is …

Proposes a new active learning approach by proposing a new uncertainty measure called the least disagree metric, as well as its efficient estimator, which is proven to be asymptotically consistent. This is then combined with seeding to become a new active learning algorith, LDM-S, which is shown to outperform existing approaches across various architectures and datasets.