Arun Verma

Postdoc@Singapore-MIT Alliance for Research and Technology·新加坡·OpenReview

研究方向

LLM Alignment and Optimization · Active Leaning · Bayesian Optimization · Reinforcement Learning · Multi-Armed Bandits · Preference-based Optimization

5.5

合作者 (16)

Bryan Kian Hsiang Low

Arun Verma

ActiveDPO: Active Direct Preference Optimization for Sample-Efficient Alignment

Provably Sample-Efficient Active Preference Data Collection

COBRA: Contextual Bandit Algorithm for Ensuring Truthful Strategic Agents

Incentivizing Time-Aware Fairness in Data Sharing

Neural Dueling Bandits: Preference-Based Optimization with Human Feedback

Prompt Optimization with Human Feedback

Achieving Exact Federated Unlearning with Improved Post-Unlearning Performance