English
Related papers

Related papers: Optimization Issues in KL-Constrained Approximate …

200 papers

Reinforcement learning (RL) has become central to enhancing reasoning in large language models (LLMs). Yet on-policy algorithms such as Group Relative Policy Optimization (GRPO) often suffer in early training: noisy gradients from…

Machine Learning · Computer Science 2026-03-19 Ziyan Wang , Zheng Wang , Xingwei Qu , Qi Cheng , Jie Fu , Shengpu Tang , Minjia Zhang , Xiaoming Huo

Aligning large language models (LLMs) on domain-specific data remains a fundamental challenge. Supervised fine-tuning (SFT) offers a straightforward way to inject domain knowledge but often degrades the model's generality. In contrast,…

Machine Learning · Computer Science 2026-02-12 Linxuan Xia , Xiaolong Yang , Yongyuan Chen , Enyue Zhao , Deng Cai , Yasheng Wang , Boxi Wu

Many advances that have improved the robustness and efficiency of deep reinforcement learning (RL) algorithms can, in one way or another, be understood as introducing additional objectives or constraints in the policy optimization step.…

Regularization is a powerful technique for extracting useful information from noisy data. Typically, it is implemented by adding some sort of norm constraint to an objective function and then exactly optimizing the modified objective…

Data Structures and Algorithms · Computer Science 2011-04-28 Michael W. Mahoney , Lorenzo Orecchia

Many policy optimization approaches in reinforcement learning incorporate a Kullback-Leilbler (KL) divergence to the previous policy, to prevent the policy from changing too quickly. This idea was initially proposed in a seminal paper on…

Machine Learning · Computer Science 2024-03-19 Lingwei Zhu , Zheng Chen , Matthew Schlegel , Martha White

The $Q$-function is a central quantity in many Reinforcement Learning (RL) algorithms for which RL agents behave following a (soft)-greedy policy w.r.t. to $Q$. It is a powerful tool that allows action selection without a model of the…

Machine Learning · Computer Science 2022-06-01 Nino Vieillard , Marcin Andrychowicz , Anton Raichuk , Olivier Pietquin , Matthieu Geist

We propose a successive convex approximation based off-policy optimization (SCAOPO) algorithm to solve the general constrained reinforcement learning problem, which is formulated as a constrained Markov decision process (CMDP) in the…

Machine Learning · Computer Science 2022-04-20 Chang Tian , An Liu , Guang Huang , Wu Luo

Deploying Reinforcement Learning (RL) agents in the real-world require that the agents satisfy safety constraints. Current RL agents explore the environment without considering these constraints, which can lead to damage to the hardware or…

Machine Learning · Computer Science 2021-03-17 Harshit Sikchi , Wenxuan Zhou , David Held

Modern policy optimization methods in reinforcement learning, such as TRPO and PPO, owe their success to the use of parameterized policies. However, while theoretical guarantees have been established for this class of algorithms, especially…

Machine Learning · Statistics 2024-02-14 Carlo Alfano , Rui Yuan , Patrick Rebeschini

Policy optimization methods with function approximation are widely used in multi-agent reinforcement learning. However, it remains elusive how to design such algorithms with statistical guarantees. Leveraging a multi-agent performance…

Machine Learning · Computer Science 2023-05-09 Yulai Zhao , Zhuoran Yang , Zhaoran Wang , Jason D. Lee

Federated Reinforcement Learning (FRL) has been deemed as a promising solution for intelligent decision-making in the era of Artificial Internet of Things. However, existing FRL approaches often entail repeated interactions with the…

Machine Learning · Computer Science 2024-05-30 Sheng Yue , Zerui Qin , Xingyuan Hua , Yongheng Deng , Ju Ren

Aligning generative models with human preference via RLHF typically suffers from overoptimization, where an imperfectly learned reward model can misguide the generative model to output undesired responses. We investigate this problem in a…

Machine Learning · Computer Science 2024-12-05 Zhihan Liu , Miao Lu , Shenao Zhang , Boyi Liu , Hongyi Guo , Yingxiang Yang , Jose Blanchet , Zhaoran Wang

In this effort, we consider the impact of regularization on the diversity of actions taken by policies generated from reinforcement learning agents trained using a policy gradient. Policy gradient agents are prone to entropy collapse, which…

Machine Learning · Computer Science 2023-10-10 Andrew Starnes , Anton Dereventsov , Clayton Webster

Direct Preference Optimization (DPO) has become a widely used training method for the instruction fine-tuning of large language models (LLMs). In this work, we explore an under-investigated aspect of DPO - its dependency on the reference…

Computation and Language · Computer Science 2024-08-23 Yixin Liu , Pengfei Liu , Arman Cohan

Reverse Kullback-Leibler (KL) divergence-based regularization with respect to a fixed reference policy is widely used in modern reinforcement learning to preserve the desired traits of the reference policy and sometimes to promote…

Machine Learning · Computer Science 2026-02-05 Anupam Nayak , Tong Yang , Osman Yagan , Gauri Joshi , Yuejie Chi

In the Bayesian reinforcement learning (RL) setting, a prior distribution over the unknown problem parameters -- the rewards and transitions -- is assumed, and a policy that optimizes the (posterior) expected return is sought. A common…

Machine Learning · Computer Science 2021-09-27 Aviv Tamar , Daniel Soudry , Ev Zisselman

One of the key challenges to deep reinforcement learning (deep RL) is to ensure safety at both training and testing phases. In this work, we propose a novel technique of unsupervised action planning to improve the safety of on-policy…

Robotics · Computer Science 2021-09-30 Hao-Lun Hsu , Qiuhua Huang , Sehoon Ha

To overcome the curses of dimensionality and modeling of Dynamic Programming (DP) methods to solve Markov Decision Process (MDP) problems, Reinforcement Learning (RL) methods are adopted in practice. Contrary to traditional RL algorithms…

Machine Learning · Computer Science 2021-08-24 Arghyadip Roy , Vivek Borkar , Abhay Karandikar , Prasanna Chaporkar

We describe an iterative procedure for optimizing policies, with guaranteed monotonic improvement. By making several approximations to the theoretically-justified procedure, we develop a practical algorithm, called Trust Region Policy…

Machine Learning · Computer Science 2017-04-24 John Schulman , Sergey Levine , Philipp Moritz , Michael I. Jordan , Pieter Abbeel

This paper proposes and analyzes two new policy learning methods: regularized policy gradient (RPG) and iterative policy optimization (IPO), for a class of discounted linear-quadratic control (LQC) problems over an infinite time horizon…

Optimization and Control · Mathematics 2025-10-08 Xin Guo , Xinyu Li , Renyuan Xu