arXiv:2607. 21999v1 Announce Type: new Abstract: Long-tailed learning couples two sources of poor generalization: head classes dominate training exposure, while under-represented classes often converge to sharper regions of the loss landscape.
arXiv:2603. 10184v2 Announce Type: replace-cross Abstract: Statistical inference with bandit data presents fundamental challenges owing to adaptive sampling, which violates the independence assumptions underlying classical asymptotic theory.
The paper introduces Fast Regularized Policy Mirror Descent (PMD) that pairs policy updates with a single temporal-difference (TD) critic step. It proves global linear convergence for finite discounted MDPs using exact coordinate-wise Bellman updates and any positive actor stepsize, regardless of critic initialization. For stochastic TD-PMD with strongly convex mirror maps, the authors achieve an expected value gap of ε after “~O(1/((1-γ)^5 σ_b ε))” transitions, without requiring trajectory resets, generative models, or nested policy evaluation loops.
arXiv:2604. 06039v2 Announce Type: replace-cross Abstract: Value iteration-type methods have been extensively studied for computing a nearly optimal value function in reinforcement learning (RL).
arXiv:2610.35551v1 Announce Type: cross Abstract: In stochastic multi-objective bandits, Pareto optimal arms may represent different trade-offs among objectives. Balanced sampling across them is impo...
arXiv:2607.02137v3 Announce Type: replace-cross Abstract: We study timestep allocation for score-based diffusion sampling, where a learned reverse-time dynamics is discretized on a finite grid. Unifo...
arXiv:2607. 02137v1 Announce Type: cross Abstract: We study timestep allocation for score-based diffusion sampling, where a learned reverse-time dynamics is discretized on a finite grid.
The paper studies distributionally robust ranking and selection (DRR&S), where the goal is to identify the best alternative under input uncertainty by considering multiple plausible input distributions. It introduces the concept of sequential additivity, showing that efficient sampling should focus on a small, additive set of critical scenarios rather than a multiplicative number. The authors prove an algorithm‑independent lower bound on sampling, design an additive allocation (AA) procedure that meets this bound and achieves exponentially decreasing error probability, and extend the approach to a general additive allocation (GAA) framework that incorporates traditional R&S sampling rules.
arXiv:2606. 27171v1 Announce Type: new Abstract: This work addresses the problem of variance in stochastic gradient estimation for machine learning optimization.
arXiv:2607. 01179v1 Announce Type: new Abstract: Scaling inference compute, by generating many parallel attempts per problem, is a costly but reliable lever for improving language model capabilities.
We study timestep allocation for score-based diffusion sampling, where a learned reverse-time dynamics is discretized on a finite grid. Uniform and hand-crafted schedules are standard choices, but they rely on fixed prescriptions and can therefore be suboptimal.
arXiv:2609.39634v1 Announce Type: cross Abstract: Common policy improvement methods, including TRPO, PPO, and GRPO, estimate policy improvement under the behavioral policy's state-visitation distribu...