14 papers
Optimal Skill Selection for LLM Agents with Provable Bicriteria Guarantees
Yu Chen, Ruishuo Chen, Xun Wang +2
Loading reusable skill documents into a bounded context window has become a primary way large language model (LLM) agents acquire task-specific capabilities, which makes skill sele…
Lipschitz Bandits with Arbitrary Feedback Delays
Yuhao Liu, Yu Chen, Longbo Huang
The Lipschitz bandit problem extends the traditional multi-armed bandit framework to continuous action spaces by assuming that the reward functions satisfy a Lipschitz condition. T…
AV-AIVAT: 74x Cheaper Agent Evaluation with Certified Anytime-Valid Stopping in Imperfect-Information Games
Boning Li, Yu Chen, Longbo Huang
Deciding which of two agents is stronger means playing games until skill outweighs luck, and every game costs money, model inference, or expert time. Since the number of games need…
Correlated Chance Sampling for Monte Carlo Counterfactual Regret Minimization
Boning Li, Yu Chen, Longbo Huang
Monte Carlo Counterfactual Regret Minimization (MCCFR) repeatedly allocates chance outcomes while its strategy evolves, yet standard sampling draws those outcomes independently on…
When Context Returns: Toward Robust Internalization in On-Policy Distillation
Xun Wang, Ruishuo Chen, Zhuoran Li +2
Recent work has shown that on-policy distillation can internalize privileged context, such as system prompts or task hints, into a student model so that the context is no longer ne…
PowerFlow: Unlocking the Dual Nature of LLMs via Principled Distribution Matching
Ruishuo Chen, Yu Chen, Zhuoran Li +1
Unsupervised Reinforcement Learning from Internal Feedback (RLIF) has emerged as a promising paradigm for eliciting the latent capabilities of Large Language Models (LLMs) without…
Best-of-Both-Worlds for Heavy-Tailed Markov Decision Processes
Yu Chen, Yuhao Liu, Jiatai Huang +2
We investigate episodic Markov Decision Processes with heavy-tailed losses (HTMDPs). Existing approaches for HTMDPs are conservative in stochastic environments and lack adaptivity…
Finite-time Convergence Analysis of Actor-Critic with Evolving Reward
Rui Hu, Yu Chen, Longbo Huang
Many popular practical reinforcement learning (RL) algorithms employ evolving reward functions-through techniques such as reward shaping, entropy regularization, or curriculum lear…
Finite-Time Convergence Analysis of ODE-based Generative Models for Stochastic Interpolants
Yuhao Liu, Rui Hu, Yu Chen +1
Stochastic interpolants offer a robust framework for continuously transforming samples between arbitrary data distributions, holding significant promise for generative modeling. De…
On the Sublinear Regret of Continuous K-Max Bandits
Yu Chen, Siwei Wang, Longbo Huang +1
The -Max combinatorial multi-armed bandit problem arises in applications such as recommendation and distributed decision making, where the reward is determined by the maximum ou…
Finite-Time Analysis of Discrete-Time Stochastic Interpolants
Yuhao Liu, Yu Chen, Rui Hu +1
The stochastic interpolant framework offers a powerful approach for constructing generative models based on ordinary differential equations (ODEs) or stochastic differential equati…
uniINF: Best-of-Both-Worlds Algorithm for Parameter-Free Heavy-Tailed MABs
Yu Chen, Jiatai Huang, Yan Dai +1
In this paper, we present a novel algorithm, uniINF, for the Heavy-Tailed Multi-Armed Bandits (HTMAB) problem, demonstrating robustness and adaptability in both stochastic and adve…
Provable Risk-Sensitive Distributional Reinforcement Learning with General Function Approximation
Yu Chen, Xiangcheng Zhang, Siwei Wang +1
In the realm of reinforcement learning (RL), accounting for risk is crucial for making decisions under uncertainty, particularly in applications where safety and reliability are pa…
Provably Efficient Partially Observable Risk-Sensitive Reinforcement Learning with Hindsight Observation
Tonghe Zhang, Yu Chen, Longbo Huang
This work pioneers regret analysis of risk-sensitive reinforcement learning in partially observable environments with hindsight observation, addressing a gap in theoretical explora…