Traditional reinforcement learning from human feedback (RLHF) approaches relying on
parametric models like the Bradley-Terry model fall short in capturing the intransitivity and …

Dueling bandits is a prominent framework for decision-making involving preferential
feedback, a valuable feature that fits various applications involving human interaction, such …

We consider the problem of contextual bandits and imitation learning, where the learner
lacks direct knowledge of the executed action's reward. Instead, the learner can actively …

Aligning large language models (LLM) with human preference plays a key role in building
modern generative models and can be achieved by reinforcement learning from human …

Learning from human feedback plays an important role in aligning generative models, such
as large language models (LLM). However, the effectiveness of this approach can be …

Reverse-Kullback-Leibler (KL) regularization has emerged to be a predominant technique
used to enhance policy optimization in reinforcement learning (RL) and reinforcement …

Contextual dueling bandits, where a learner compares two options based on context and
receives feedback indicating which was preferred, extends classic dueling bandits by …

Bandit is acknowledged as a classical analytic tool for the online decision-making problem
under uncertainty, eg, task assignment for crowdsourcing systems given the unknown …

This dissertation investigates several challenges in artificial intelligence (AI) alignment and
reinforcement learning (RL), particularly focusing on applications when only preference …