在众多慢性病管理和重症监护应用中，人们对采用数据驱动方法推荐最优治疗策略的兴趣日益浓厚。强化学习方法非常适合这类序贯决策问题，但由于直接在线探索存在安全风险且不可行，因此必须完全基于回顾性医疗记录数据集进行训练和评估。尽管有这一要求，绝大多数治疗优化研究仍采用离策略RL方法（例如Double Deep Q Networks（DDQN，双深度Q网络）及其变体），而这类方法在纯离线场景下的表现众所周知很差。离线RL领域的最新进展，例如Conservative Q-Learning（CQL，保守Q学习），提供了合适的替代方案。但要将这些方法适配到真实世界应用中仍存在挑战，这类应用里回顾性数据集中次优样本占主导，且需要满足严格的安全约束。在本研究中，我们提出了一种实用且有理论依据的转移采样方法，用于解决离线RL训练过程中的动作不平衡问题。我们在糖尿病和脓毒症治疗优化这两个真实世界任务上开展了大量实验，将所提方法的性能与主流的离策略及离线RL基准（DDQN和CQL）进行对比。在一系列符合原则且具有临床相关性的指标上，我们的研究表明，所提方法能大幅改善预期健康结果，且符合相关临床实践与安全指南。
英文摘要：
There is increasing interest in data-driven approaches for recommending optimal treatment strategies in many chronic disease management and critical care applications. Reinforcement learning methods are well-suited to this sequential decision-making problem, but must be trained and evaluated exclusively on retrospective medical record datasets as direct online exploration is unsafe and infeasible. Despite this requirement, the vast majority of treatment optimization studies use off-policy RL methods (e.g., Double Deep Q Networks (DDQN) or its variants) that are known to perform poorly in purely offline settings. Recent advances in offline RL, such as Conservative Q-Learning (CQL), offer a suitable alternative. But there remain challenges in adapting these approaches to real-world applications where suboptimal examples dominate the retrospective dataset and strict safety constraints need to be satisfied. In this work, we introduce a practical and theoretically grounded transition sampling approach to address action imbalance during offline RL training. We perform extensive experiments on two real-world tasks for diabetes and sepsis treatment optimization to compare performance of the proposed approach against prominent off-policy and offline RL baselines (DDQN and CQL). Across a range of principled and clinically relevant metrics, we show that our proposed approach enables substantial improvements in expected health outcomes and in accordance with relevant practice and safety guidelines.
L3S Research Center（L3S研究中心）；Leibniz University Hannover（汉诺威莱布尼茨大学）
The success of deep reinforcement learning (DRL) hinges on the availability of training data, which is typically obtained via a large number of environment interactions. In many real-world scenarios, costs and risks are associated with gathering these data. The field of offline reinforcement learning addresses these issues through outsourcing the collection of data to a domain expert or a carefully monitored program and subsequently searching for a batch-constrained optimal policy. With the emergence of data markets, an alternative to constructing a dataset in-house is to purchase external dat…
Nanjing University（南京大学）；The University of Hong Kong（香港大学）
Value function estimation is an indispensable subroutine in reinforcement learning, which becomes more challenging in the offline setting. In this paper, we propose Hybrid Value Estimation (HVE) to reduce value estimation error, which trades off bias and variance by balancing between the value estimation from offline data and the learned model. Theoretical analysis discloses that HVE enjoys a better error bound than the direct methods. HVE can be leveraged in both off-policy evaluation and offline reinforcement learning settings. We, therefore, provide two concrete algorithms Off-policy HVE (O…
Institute for Interdisciplinary Information Sciences, Tsinghua University（清华大学交叉信息研究院）；Huawei Noah’s Ark Lab（华为诺亚方舟实验室）
With the success of offline reinforcement learning (RL), offline trained RL policies have the potential to be further improved when deployed online. A smooth transfer of the policy matters in safe real-world deployment. Besides, fast adaptation of the policy plays a vital role in practical online performance improvement. To tackle these challenges, we propose a simple yet efficient algorithm, Model-based Offline-to-Online Reinforcement learning (MOORe), which employs a prioritized sampling scheme that can dynamically adjust the offline and online data for smooth and efficient online adaptation…
Siemens Technology（西门子技术）；Technical University of Munich（慕尼黑工业大学）
In offline reinforcement learning, a policy needs to be learned from a single pre-collected dataset. Typically, policies are thus regularized during training to behave similarly to the data generating policy, by adding a penalty based on a divergence between action distributions of generating and trained policy. We propose a new algorithm, which constrains the policy directly in its weight space instead, and demonstrate its effectiveness in experiments.
Offline reinforcement learning (RL) allows for the training of competent agents from offline datasets without any interaction with the environment. Online finetuning of such offline models can further improve performance. But how should we ideally finetune agents obtained from offline RL training? While offline RL algorithms can in principle be used for finetuning, in practice, their online performance improves slowly. In contrast, we show that it is possible to use standard online off-policy algorithms for faster improvement. However, we find this approach may suffer from policy collapse, whe…
Westlake University（西湖大学）；Zhejiang University（浙江大学）；Westlake Institute for Advanced Study（西湖高等研究院）
Offline reinforcement learning algorithms promise to be applicable in settings where a fixed dataset is available and no new experience can be acquired. However, such formulation is inevitably offline-data-hungry and, in practice, collecting a large offline dataset for one specific task over one specific environment is also costly and laborious. In this paper, we thus 1) formulate the offline dynamics adaptation by using (source) offline data collected from another dynamics to relax the requirement for the extensive (target) offline data, 2) characterize the dynamics shift problem in which pri…
Siemens Technology（西门子技术部）；Technical University of Munich（慕尼黑工业大学）
Offline reinforcement learning (RL) Algorithms are often designed with environments such as MuJoCo in mind, in which the planning horizon is extremely long and no noise exists. We compare model-free, model-based, as well as hybrid offline RL approaches on various industrial benchmark (IB) datasets to test the algorithms in settings closer to real world problems, including complex noise and partially observable states. We find that on the IB, hybrid approaches face severe difficulties and that simpler algorithms, such as rollout based algorithms or model-free algorithms with simpler regularizer…
Learning efficiently from small amounts of data has long been the focus of model-based reinforcement learning, both for the online case when interacting with the environment and the offline case when learning from a fixed dataset. However, to date no single unified algorithm could demonstrate state-of-the-art results in both settings. In this work, we describe the Reanalyse algorithm which uses model-based policy and value improvement operators to compute new improved training targets on existing data points, allowing efficient learning for data budgets varying by several orders of magnitude. …
Training Reinforcement Learning (RL) agents in high-stakes applications might be too prohibitive due to the risk associated to exploration. Thus, the agent can only use data previously collected by safe policies. While previous work considers optimizing the average performance using offline data, we focus on optimizing a risk-averse criteria, namely the CVaR. In particular, we present the Offline Risk-Averse Actor-Critic (O-RAAC), a model-free RL algorithm that is able to learn risk-averse policies in a fully offline setting. We show that O-RAAC learns policies with higher CVaR than risk-neutr…
Sufficient exploration is paramount for the success of a reinforcement learning agent. Yet, exploration is rarely assessed in an algorithm-independent way. We compare the behavior of three data-based, offline exploration metrics described in the literature on intuitive simple distributions and highlight problems to be aware of when using them. We propose a fourth metric,uniform relative entropy, and implement it using either a k-nearest-neighbor or a nearest-neighbor-ratio estimator, highlighting that the implementation choices have a profound impact on these measures.
We present a model-based offline reinforcement learning policy performance lower bound that explicitly captures dynamics model misspecification and distribution mismatch and we propose an empirical algorithm for optimal offline policy selection. Theoretically, we prove a novel safe policy improvement theorem by establishing pessimism approximations to the value function. Our key insight is to jointly consider selecting over dynamics models and policies: as long as a dynamics model can accurately represent the dynamics of the state-action pairs visited by a given policy, it is possible to appro…
Offline reinforcement learning (RL) enables effective learning from previously collected data without exploration, which shows great promise in real-world applications when exploration is expensive or even infeasible. The discount factor, $γ$, plays a vital role in improving online RL sample efficiency and estimation accuracy, but the role of the discount factor in offline RL is not well explored. This paper examines two distinct effects of $γ$ in offline RL with theoretical analysis, namely the regularization effect and the pessimism effect. On the one hand, $γ$ is a regulator to trade-off op…
New York University（纽约大学）；Google Research（谷歌研究院）；Google DeepMind（谷歌DeepMind）
Many modern approaches to offline Reinforcement Learning (RL) utilize behavior regularization, typically augmenting a model-free actor critic algorithm with a penalty measuring divergence of the policy from the offline data. In this work, we propose an alternative approach to encouraging the learned policy to stay close to the data, namely parameterizing the critic as the log-behavior-policy, which generated the offline data, plus a state-action value offset term, which can be learned using a neural network. Behavior regularization then corresponds to an appropriate regularizer on the offset t…
Offline reinforcement learning algorithms still lack trust in practice due to the risk that the learned policy performs worse than the original policy that generated the dataset or behaves in an unexpected way that is unfamiliar to the user. At the same time, offline RL algorithms are not able to tune their most important hyperparameter - the proximity of the learned policy to the original policy. We propose an algorithm that allows the user to tune this hyperparameter at runtime, thereby addressing both of the above mentioned issues simultaneously. This allows users to start with the original…
Google Brain（谷歌大脑）；UC Berkeley（加州大学伯克利分校）；MIT（麻省理工学院）；Google Research（谷歌研究院）
Reinforcement learning (RL) has achieved impressive performance in a variety of online settings in which an agent's ability to query the environment for transitions and rewards is effectively unlimited. However, in many practical applications, the situation is reversed: an agent may have access to large amounts of undirected offline experience data, while access to the online environment is severely limited. In this work, we focus on this offline setting. Our main insight is that, when presented with offline data composed of a variety of behaviors, an effective way to leverage this data is to …
In offline reinforcement learning, a policy learns to maximize cumulative rewards with a fixed collection of data. Towards conservative strategy, current methods choose to regularize the behavior policy or learn a lower bound of the value function. However, exorbitant conservation tends to impair the policy's generalization ability and degrade its performance, especially for the mixed datasets. In this paper, we propose the method of reducing conservativeness oriented reinforcement learning. On the one hand, the policy is trained to pay more attention to the minority samples in the static data…
Many practical applications of reinforcement learning (RL) constrain the agent to learn from a fixed offline dataset of logged interactions, which has already been gathered, without offering further possibility for data collection. However, commonly used off-policy RL algorithms, such as the Deep Q Network and the Deep Deterministic Policy Gradient, are incapable of learning without data correlated to the distribution under the current policy, making them ineffective for this offline setting. As the first step towards useful offline RL algorithms, we analysis the reason of instability in stand…
Mobile notification systems have taken a major role in driving and maintaining user engagement for online platforms. They are interesting recommender systems to machine learning practitioners with more sequential and long-term feedback considerations. Most machine learning applications in notification systems are built around response-prediction models, trying to attribute both short-term impact and long-term impact to a notification decision. However, a user's experience depends on a sequence of notifications and attributing impact to a single notification is not always accurate, if not impos…
University of California, Berkeley（加州大学伯克利分校）；DeepMind（深度思维）
Offline Reinforcement Learning (ORL) enablesus to separately study the two interlinked processes of reinforcement learning: collecting informative experience and inferring optimal behaviour. The second step has been widely studied in the offline setting, but just as critical to data-efficient RL is the collection of informative data. The task-agnostic setting for data collection, where the task is not known a priori, is of particular interest due to the possibility of collecting a single dataset and using it to solve several downstream tasks as they arise. We investigate this setting via curio…
Efficient and Uncertainty-Aware Diffusion Framework for Offline-to-Online Reinforcement Learning
Department of Computer Science, Johns Hopkins University, Baltimore, MD, U.S.A.（约翰霍普金斯大学计算机科学系）；AMLab, University of Amsterdam, Amsterdam, Netherlands（阿姆斯特丹大学AM实验室）
Offline Reinforcement Learning (RL) aims to turn large datasets into powerful decision-making engines without any online interactions with the environment. This great promise has motivated a large amount of research that hopes to replicate the success RL has experienced in simulation settings. This work ambitions to reflect upon these efforts from a practitioner viewpoint. We start by discussing the dataset properties that we hypothesise can characterise the type of offline methods that will be the most successful. We then verify these claims through a set of experiments and designed datasets …
Many reinforcement learning (RL) problems in practice are offline, learning purely from observational data. A key challenge is how to ensure the learned policy is safe, which requires quantifying the risk associated with different actions. In the online setting, distributional RL algorithms do so by learning the distribution over returns (i.e., cumulative rewards) instead of the expected return; beyond quantifying risk, they have also been shown to learn better representations for planning. We propose Conservative Offline Distributional Actor Critic (CODAC), an offline RL algorithm suitable fo…
Tsinghua Shenzhen International Graduate School（清华大学深圳国际研究生院）
Offline Reinforcement Learning has attracted much interest in solving the application challenge for traditional reinforcement learning. Offline reinforcement learning uses previously-collected datasets to train agents without any interaction. For addressing the overestimation of OOD (out-of-distribution) actions, conservative estimates give a low value for all inputs. Previous conservative estimation methods are usually difficult to avoid the impact of OOD actions on Q-value estimates. In addition, these algorithms usually need to lose some computational efficiency to achieve the purpose of co…
Siemens Technology（西门子技术）；Technical University of Munich（慕尼黑工业大学）
Recently developed offline reinforcement learning algorithms have made it possible to learn policies directly from pre-collected datasets, giving rise to a new dilemma for practitioners: Since the performance the algorithms are able to deliver depends greatly on the dataset that is presented to them, practitioners need to pick the right dataset among the available ones. This problem has so far not been discussed in the corresponding literature. We discuss ideas how to select promising datasets and propose three very simple indicators: Estimated relative return improvement (ERI) and estimated a…
Apple Inc.（苹果公司）；Carnegie Mellon University（卡内基梅隆大学）
Offline Reinforcement Learning promises to learn effective policies from previously-collected, static datasets without the need for exploration. However, existing Q-learning and actor-critic based off-policy RL algorithms fail when bootstrapping from out-of-distribution (OOD) actions or states. We hypothesize that a key missing ingredient from the existing methods is a proper treatment of uncertainty in the offline setting. We propose Uncertainty Weighted Actor-Critic (UWAC), an algorithm that detects OOD state-action pairs and down-weights their contribution in the training objectives accordi…
Conservatism has led to significant progress in offline reinforcement learning (RL) where an agent learns from pre-collected datasets. However, as many real-world scenarios involve interaction among multiple agents, it is important to resolve offline RL in the multi-agent setting. Given the recent success of transferring online RL algorithms to the multi-agent setting, one may expect that offline RL algorithms will also transfer to multi-agent settings directly. Surprisingly, we empirically observe that conservative offline RL algorithms do not work well in the multi-agent setting -- the perfo…
Tokyo Institute of Technology（东京工业大学）；negocia, Inc.（Negocia公司）；Hanjuku-Kaso Co., Ltd.（Hanjuku-Kaso有限公司）；Cornell University（康奈尔大学）
In recommender systems (RecSys) and real-time bidding (RTB) for online advertisements, we often try to optimize sequential decision making using bandit and reinforcement learning (RL) techniques. In these applications, offline reinforcement learning (offline RL) and off-policy evaluation (OPE) are beneficial because they enable safe policy optimization using only logged data without any risky online interaction. In this position paper, we explore the potential of using simulation to accelerate practical research of offline RL and OPE, particularly in RecSys and RTB. Specifically, we discuss ho…
We hypothesize that empirically studying the sample complexity of offline reinforcement learning (RL) is crucial for the practical applications of RL in the real world. Several recent works have demonstrated the ability to learn policies directly from offline data. In this work, we ask the question of the dependency on the number of samples for learning from offline data. Our objective is to emphasize that studying sample complexity for offline RL is important, and is an indicator of the usefulness of existing offline algorithms. We propose an evaluation approach for sample complexity analysis…
Recent advances in batch (offline) reinforcement learning have shown promising results in learning from available offline data and proved offline reinforcement learning to be an essential toolkit in learning control policies in a model-free setting. An offline reinforcement learning algorithm applied to a dataset collected by a suboptimal non-learning-based algorithm can result in a policy that outperforms the behavior agent used to collect the data. Such a scenario is frequent in robotics, where existing automation is collecting operational data. Although offline learning techniques can learn…
King Abdullah University of Science and Technology（阿卜杜拉国王科技大学）；The Swiss AI Lab IDSIA/USI/SUPSI（瑞士人工智能实验室 IDSIA/USI/SUPSI）；NNAISENSE
Offline RL methods have been shown to reduce the need for environment interaction by training agents using offline collected episodes. However, these methods typically require action information to be logged during data collection, which can be difficult or even impossible in some practical cases. In this paper, we investigate the potential of using action-free offline datasets to improve online reinforcement learning, name this problem Reinforcement Learning with Action-Free Offline Pretraining (AFP-RL). We introduce Action-Free Guide (AF-Guide), a method that guides online training by extrac…
Offline goal-conditioned reinforcement learning (GCRL) can be challenging due to overfitting to the given dataset. To generalize agents' skills outside the given dataset, we propose a goal-swapping procedure that generates additional trajectories. To alleviate the problem of noise and extrapolation errors, we present a general offline reinforcement learning method called deterministic Q-advantage policy gradient (DQAPG). In the experiments, DQAPG outperforms state-of-the-art goal-conditioned offline RL methods in a wide range of benchmark tasks, and goal-swapping further improves the test resu…
