论文检索

Reinforcement Learning · Deep RL

Lior Shani, Aviv Rosenberg, Asaf Cassel, Oran Lang, Daniele Calandriello, Avital Zipori, Hila Noga, Orgad Keller, Bilal Piot, Idan Szpektor 等

Reinforcement Learning from Human Feedback (RLHF) has become the standard approach for aligning Large Language Models (LLMs) with human preferences, allowing LLMs to demonstrate remarkable abilities in various tasks. Existing methods work by emulating the human preference at the single decision (turn) level, limiting their capabilities in settings that require planning or multi-turn interactions to achieve a long-term goal. In this paper, we address this issue by developing novel methods for Reinforcement Learning (RL) from preference feedback between two full multi-turn conversations. In the tabular setting, we present a novel mirror-descent-based policy optimization algorithm for the general multi-turn preference-based RL problem, and prove its convergence to Nash equilibrium. To evaluate performance, we create a new environment, Education Dialogue, where a teacher agent guides a student in learning a random topic, and show that a deep RL variant of our algorithm outperforms RLHF baselines. Finally, we show that in an environment with explicit rewards, our algorithm recovers the same performance as a reward-based RL baseline, despite relying solely on a weaker preference signal.

Theory · Reinforcement Learning and Planning

Jincheng Mei, Bo Dai, Alekh Agarwal, Sharan Vaswani, Anant Raj, Csaba Szepesvari, Dale Schuurmans

We provide a new understanding of the stochastic gradient bandit algorithm by showing that it converges to a globally optimal policy almost surely using \emph{any} constant learning rate. This result demonstrates that the stochastic gradient algorithm continues to balance exploration and exploitation appropriately even in scenarios where standard smoothness and noise control assumptions break down. The proofs are based on novel findings about action sampling rates and the relationship between cumulative progress and noise, and extend the current understanding of how simple stochastic gradient methods behave in bandit settings.

Theory · Reinforcement Learning and Planning

Tyler Sam, Yudong Chen, Christina Yu

Many reinforcement learning (RL) algorithms are too costly to use in practice due to the large sizes $S,A$ of the problem's state and action space. To resolve this issue, we study transfer RL with latent low rank structure. We consider the problem of transferring a latent low rank representation when the source and target MDPs have transition kernels with Tucker rank $(S, d, A)$, $(S ,S , d), (d, S , A )$, or $(d , d , d )$. In each setting, we introduce the transfer-ability coefficient $\alpha$ that measures the difficulty of representational transfer. Our algorithm learns latent representations in each source MDP and then exploits the linear structure to remove the dependence on $S , A $, or $SA $ in the target MDP regret bound. We complement our positive results with information theoretic lower bounds that show our algorithms (excluding the ($d, d, d$) setting) are minimax-optimal with respect to $\alpha$.

Generative Models

Yujia Zhou, Zheng Liu, Zhicheng Dou

The emergence of Large Language Models (LLMs) has significantly advanced natural language processing, but these models often generate factually incorrect information, known as "hallucination." Initial retrieval-augmented generation (RAG) methods like the "Retrieve-Read" framework was inadequate for complex reasoning tasks. Subsequent prompt-based RAG strategies and Supervised Fine-Tuning (SFT) methods improved performance but required frequent retraining and risked altering foundational LLM capabilities. To cope with these challenges, we propose Assistant-based Retrieval-Augmented Generation (AssistRAG), integrating an intelligent information assistant within LLMs. This assistant manages memory and knowledge through tool usage, action execution, memory building, and plan specification. Using a two-phase training approach—Curriculum Assistant Learning and Reinforced Preference Optimization—AssistRAG enhances information retrieval and decision-making. Experiments show AssistRAG significantly outperforms benchmarks, especially benefiting less advanced LLMs, by providing superior reasoning capabilities and accurate responses.

Reinforcement Learning · Planning

Ye Tian, Baolin Peng, Linfeng Song, Lifeng Jin, Dian Yu, Lei Han, Haitao Mi, Dong Yu

Despite the impressive capabilities of Large Language Models (LLMs) on various tasks, they still struggle with scenarios that involves complex reasoning and planning. Self-correction and self-learning emerge as viable solutions, employing strategies that allow LLMs to refine their outputs and learn from self-assessed rewards. Yet, the efficacy of LLMs in self-refining its response, particularly in complex reasoning and planning task, remains dubious. In this paper, we introduce AlphaLLM for the self-improvements of LLMs, which integrates Monte Carlo Tree Search (MCTS) with LLMs to establish a self-improving loop, thereby enhancing the capabilities of LLMs without additional annotations. Drawing inspiration from the success of AlphaGo, AlphaLLM addresses the unique challenges of combining MCTS with LLM for self-improvement, including data scarcity, the vastness search spaces of language tasks, and the subjective nature of feedback in language tasks. AlphaLLM is comprised of prompt synthesis component, an efficient MCTS approach tailored for language tasks, and a trio of critic models for precise feedback. Our experimental results in mathematical reasoning tasks demonstrate that AlphaLLM significantly enhances the performance of LLMs without additional annotations, showing the potential for self-improvement in LLMs.

Language

Zichao Li, Yanshuai Cao, Jackie CK Cheung

How do large language models (LLMs) encode the state of the world, including the status of entities and their relations, as described by a text? While existing work directly probes for a complete state of the world, our research explores whether and how LLMs abstract this world state in their internal representations. We propose a new framework for probing for world representations through the lens of state abstraction theory from reinforcement learning, which emphasizes different levels of abstraction, distinguishing between general abstractions that facilitate predicting future states and goal-oriented abstractions that guide the subsequent actions to accomplish tasks. To instantiate this framework, we design a text-based planning task, where an LLM acts as an agent in an environment and interacts with objects in containers to achieve a specified goal state. Our experiments reveal that fine-tuning as well as advanced pre-training strengthens LLM-built representations' tendency of maintaining goal-oriented abstractions during decoding, prioritizing task completion over recovery of the world's state and dynamics.

Reinforcement Learning · Planning

Michael Katz, Harsha Kokel, Kavitha Srinivas, Shirin Sohrabi Araghi

Among the most important properties of algorithms investigated in computer science are soundness, completeness, and complexity. These properties, however, are rarely analyzed for the vast collection of recently proposed methods for planning with large language models. In this work, we alleviate this gap. We analyse these properties of using LLMs for planning and highlight that recent trends abandon both soundness and completeness for the sake of inefficiency. We propose a significantly more efficient approach that can, at the same time, maintain both soundness and completeness. We exemplify on four representative search problems, comparing to the LLM-based solutions from the literature that attempt to solve these problems. We show that by using LLMs to produce the code for the search components we can solve the entire datasets with 100% accuracy with only a few calls to the LLM. In contrast, the compared approaches require hundreds of thousands of calls and achieve significantly lower accuracy. We argue for a responsible use of compute resources; urging research community to investigate sound and complete LLM-based approaches that uphold efficiency.

Theory · Reinforcement Learning and Planning

Pietro Novelli, Marco Pratticò, Massimiliano Pontil, Carlo Ciliberto

Policy Mirror Descent (PMD) is a powerful and theoretically sound methodology for sequential decision-making. However, it is not directly applicable to Reinforcement Learning (RL) due to the inaccessibility of explicit action-value functions. We address this challenge by introducing a novel approach based on learning a world model of the environment using conditional mean embeddings. Leveraging tools from operator theory we derive a closed-form expression of the action-value function in terms of the world model via simple matrix operations. Combining these estimators with PMD leads to POWR, a new RL algorithm for which we prove convergence rates to the global optimum. Preliminary experiments in finite and infinite state settings support the effectiveness of our method.

Theory · Reinforcement Learning and Planning

Xuan Bach Le, Dominik Wagner, Leon Witzman, Alexander Rabinovich, Luke Ong

Linear temporal logic (LTL) and, more generally, $\omega$-regular objectives are alternatives to the traditional discount sum and average reward objectives in reinforcement learning (RL), offering the advantage of greater comprehensibility and hence explainability. In this work, we study the relationship between these objectives. Our main result is that each RL problem for $\omega$-regular objectives can be reduced to a limit-average reward problem in an optimality-preserving fashion, via (finite-memory) reward machines. Furthermore, we demonstrate the efficacy of this approach by showing that optimal policies for limit-average problems can be found asymptotically by solving a sequence of discount-sum problems approximately. Consequently, we resolve an open problem: optimal policies for LTL and $\omega$-regular objectives can be learned asymptotically.

Reinforcement Learning · Planning

Zizhao Wang, Jiaheng Hu, Caleb Chuck, Stephen Chen, Roberto Martín-Martín, Amy Zhang, Scott Niekum, Peter Stone

Unsupervised skill discovery carries the promise that an intelligent agent can learn reusable skills through autonomous, reward-free interactions with environments. Existing unsupervised skill discovery methods learn skills by encouraging distinguishable behaviors that cover diverse states. However, in complex environments with many state factors (e.g., household environments with many objects), learning skills that cover all possible states is impossible, and naively encouraging state diversity often leads to simple skills that are not ideal for solving downstream tasks. This work introduces Skill Discovery from Local Dependencies (SkiLD), which leverages state factorization as a natural inductive bias to guide the skill learning process. The key intuition guiding SkiLD is that skills that induce \textbf{diverse interactions} between state factors are often more valuable for solving downstream tasks. To this end, SkiLD develops a novel skill learning objective that explicitly encourages the mastering of skills that effectively induce different interactions within an environment. We evaluate SkiLD in several domains with challenging, long-horizon sparse reward tasks including a realistic simulated household robot domain, where SkiLD successfully learns skills with clear semantic meaning and shows superior performance compared to existing unsupervised reinforcement learning methods that only maximize state coverage.

Reinforcement Learning · Deep RL

Miles Hutson, Isaac Kauvar, Nick Haber

Model-based reinforcement learning (MBRL) is a promising route to sample-efficient policy optimization. However, a known vulnerability of reconstruction-based MBRL consists of scenarios in which detailed aspects of the world are highly predictable, but irrelevant to learning a good policy. Such scenarios can lead the model to exhaust its capacity on meaningless content, at the cost of neglecting important environment dynamics. While existing approaches attempt to solve this problem, we highlight its continuing impact on leading MBRL methods ---including DreamerV3 and DreamerPro--- with a novel environment where background distractions are intricate, predictable, and useless for planning future actions. To address this challenge we develop a method for focusing the capacity of the world model through a synergy of a pretrained segmentation model, a task-aware reconstruction loss, and adversarial learning. Our method outperforms a variety of other approaches designed to reduce the impact of distractors, and is an advance towards robust model-based reinforcement learning.

Optimization · Optimization and Learning under Uncertainty

Jose Pablo Folch, Calvin Tsay, Robert Lee, Behrang Shafei, Weronika Ormaniec, Andreas Krause, Mark van der Wilk, Ruth Misener, Mojmir Mutny

Bayesian optimization is a methodology to optimize black-box functions. Traditionally, it focuses on the setting where you can arbitrarily query the search space. However, many real-life problems do not offer this flexibility; in particular, the search space of the next query may depend on previous ones. Example challenges arise in the physical sciences in the form of local movement constraints, required monotonicity in certain variables, and transitions influencing the accuracy of measurements. Altogether, such *transition constraints* necessitate a form of planning. This work extends classical Bayesian optimization via the framework of Markov Decision Processes. We iteratively solve a tractable linearization of our utility function using reinforcement learning to obtain a policy that plans ahead for the entire horizon. This is a parallel to the optimization of an *acquisition function in policy space*. The resulting policy is potentially history-dependent and non-Markovian. We showcase applications in chemical reactor optimization, informative path planning, machine calibration, and other synthetic examples.

Theory · Reinforcement Learning and Planning

Nadav Merlis

We study reinforcement learning (RL) problems in which agents observe the reward or transition realizations at their current state _before deciding which action to take_. Such observations are available in many applications, including transactions, navigation and more. When the environment is known, previous work shows that this lookahead information can drastically increase the collected reward. However, outside of specific applications, existing approaches for interacting with unknown environments are not well-adapted to these observations. In this work, we close this gap and design provably-efficient learning algorithms able to incorporate lookahead information. To achieve this, we perform planning using the empirical distribution of the reward and transition observations, in contrast to vanilla approaches that only rely on estimated expectations. We prove that our algorithms achieve tight regret versus a baseline that also has access to lookahead information -- linearly increasing the amount of collected reward compared to agents that cannot handle lookahead information.

Reinforcement Learning · Planning

Stephen Chung, Scott Niekum, David Krueger

As reinforcement learning agents become increasingly deployed in real-world scenarios, predicting future agent actions and events during deployment is important for facilitating better human-agent interaction and preventing catastrophic outcomes. This paper experimentally evaluates and compares the effectiveness of future action and event prediction for three types of RL agents: explicitly planning, implicitly planning, and non-planning. We employ two approaches: the inner state approach, which involves predicting based on the inner computations of the agents (e.g., plans or neuron activations), and a simulation-based approach, which involves unrolling the agent in a learned world model. Our results show that the plans of explicitly planning agents are significantly more informative for prediction than the neuron activations of the other types. Furthermore, using internal plans proves more robust to model quality compared to simulation-based approaches when predicting actions, while the results for event prediction are more mixed. These findings highlight the benefits of leveraging inner states and simulations to predict future agent actions and events, thereby improving interaction and safety in real-world deployments.

Theory · Reinforcement Learning and Planning

Yuheng Zhang, Nan Jiang

We study off-policy evaluation (OPE) in partially observable environments with complex observations, with the goal of developing estimators whose guarantee avoids exponential dependence on the horizon. While such estimators exist for MDPs and POMDPs can be converted to history-based MDPs, their estimation errors depend on the state-density ratio for MDPs which becomes history ratios after conversion, an exponential object. Recently, Uehara et al. [2022a] proposed future-dependent value functions as a promising framework to address this issue, where the guarantee for memoryless policies depends on the density ratio over the latent state space. However, it also depends on the boundedness of the future-dependent value function and other related quantities, which we show could be exponential-in-length and thus erasing the advantage of the method. In this paper, we discover novel coverage assumptions tailored to the structure of POMDPs, such as outcome coverage and belief coverage, which enable polynomial bounds on the aforementioned quantities. As a side product, our analyses also lead to the discovery of new algorithms with complementary properties.

Theory · Reinforcement Learning and Planning

Sergio Calo, Anders Jonsson, Gergely Neu, Ludovic Schwartz, Javier Segovia-Aguas

We propose a new framework for formulating optimal transport distances between Markov chains. Previously known formulations studied couplings between the entire joint distribution induced by the chains, and derived solutions via a reduction to dynamic programming (DP) in an appropriately defined Markov decision process. This formulation has, however, not led to particularly efficient algorithms so far, since computing the associated DP operators requires fully solving a static optimal transport problem, and these operators need to be applied numerous times during the overall optimization process. In this work, we develop an alternative perspective by considering couplings between a ``flattened'' version of the joint distributions that we call discounted occupancy couplings, and show that calculating optimal transport distances in the full space of joint distributions can be equivalently formulated as solving a linear program (LP) in this reduced space. This LP formulation formulation allows us to port several algorithmic ideas from other areas of optimal transport theory. In particular, our formulation makes it possible to introduce an appropriate notion of entropy regularization into the optimization problem, which in turn enables us to directly calculate optimal transport distances via a Sinkhorn-like method we call Sinkhorn Value Iteration (SVI). We show both theoretically and empirically that this method converges quickly to an optimal coupling, essentially at the same computational cost of running vanilla Sinkhorn in each pair of states. Along the way, we point out that our optimal transport distance exactly matches the common notion of bisimulation metrics between Markov chains, and thus our results also apply to computing such metrics, and in fact our algorithm turns out to be significantly more efficient than the best known methods developed so far for this purpose.

Theory · Reinforcement Learning and Planning

Andreas Schlaginhaufen, Maryam Kamgarpour

Inverse reinforcement learning (IRL) aims to infer a reward from expert demonstrations, motivated by the idea that the reward, rather than the policy, is the most succinct and transferable description of a task [Ng et al., 2000]. However, the reward corresponding to an optimal policy is not unique, making it unclear if an IRL-learned reward is transferable to new transition laws in the sense that its optimal policy aligns with the optimal policy corresponding to the expert's true reward. Past work has addressed this problem only under the assumption of full access to the expert's policy, guaranteeing transferability when learning from two experts with the same reward but different transition laws that satisfy a specific rank condition [Rolland et al., 2022]. In this work, we show that the conditions developed under full access to the expert's policy cannot guarantee transferability in the more practical scenario where we have access only to demonstrations of the expert. Instead of a binary rank condition, we propose principal angles as a more refined measure of similarity and dissimilarity between transition laws. Based on this, we then establish two key results: 1) a sufficient condition for transferability to any transition laws when learning from at least two experts with sufficiently different transition laws, and 2) a sufficient condition for transferability to local changes in the transition law when learning from a single expert. Furthermore, we also provide a probably approximately correct (PAC) algorithm and an end-to-end analysis for learning transferable rewards from demonstrations of multiple experts.

Reinforcement Learning · Everything Else

Shaoteng Liu, Haoqi Yuan, Minda Hu, Yanwei Li, Yukang Chen, Shu Liu, Zongqing Lu, Jiaya Jia

Large Language Models (LLMs) have demonstrated proficiency in utilizing various tools by coding, yet they face limitations in handling intricate logic and precise control. In embodied tasks, high-level planning is amenable to direct coding, while low-level actions often necessitate task-specific refinement, such as Reinforcement Learning (RL). To seamlessly integrate both modalities, we introduce a two-level hierarchical framework, RL-GPT, comprising a slow agent and a fast agent. The slow agent analyzes actions suitable for coding, while the fast agent executes coding tasks. This decomposition effectively focuses each agent on specific tasks, proving highly efficient within our pipeline. Our approach outperforms traditional RL methods and existing GPT agents, demonstrating superior efficiency. In the Minecraft game, it rapidly obtains diamonds within a single day on an RTX3090. Additionally, it achieves SOTA performance across all designated MineDojo tasks.

Theory · Reinforcement Learning and Planning

Gang Ding, Zeyuan Liu, Zhirui Fang, Kefan Su, Liwen Zhu, Zongqing Lu

The partial observability and stochasticity in multi-agent settings can be mitigated by accessing more information about others via communication. However, the coordination problem still exists since agents cannot communicate actual actions with each other at the same time due to the circular dependencies. In this paper, we propose a novel multi-level communication scheme, Sequential Communication (SeqComm). SeqComm treats agents asynchronously (the upper-level agents make decisions before the lower-level ones) and has two communication phases. In the negotiation phase, agents determine the priority of decision-making by communicating hidden states of observations and comparing the value of intention, which is obtained by modeling the environment dynamics. In the launching phase, the upper-level agents take the lead in making decisions and then communicate their actions with the lower-level agents. Theoretically, we prove the policies learned by SeqComm are guaranteed to improve monotonically and converge. Empirically, we show that SeqComm outperforms existing methods in a variety of cooperative multi-agent tasks.

Reinforcement Learning · Planning

Dengwei Zhao, Shikui Tu, Lei Xu

Monte-Carlo tree search (MCTS) and reinforcement learning contributed crucially to the success of AlphaGo and AlphaZero, and A$^*$ is a tree search algorithm among the most well-known ones in the classical AI literature. MCTS and A$^*$ both perform heuristic search and are mutually beneficial. Efforts have been made to the renaissance of A$^*$ from three possible aspects, two of which have been confirmed by studies in recent years, while the third is about the OPEN list that consists of open nodes of A$^*$ search, but still lacks deep investigation. This paper aims at the third, i.e., developing the Sampling-exploration enhanced A$^*$ (SeeA$^*$) search by constructing a dynamic subset of OPEN through a selective sampling process, such that the node with the best heuristic value in this subset instead of in the OPEN is expanded. Nodes with the best heuristic values in OPEN are most probably picked into this subset, but sometimes may not be included, which enables SeeA$^*$ to explore other promising branches. Three sampling techniques are presented for comparative investigations. Moreover, under the assumption about the distribution of prediction errors, we have theoretically shown the superior efficiency of SeeA$^*$ over A$^*$ search, particularly when the accuracy of the guiding heuristic function is insufficient. Experimental results on retrosynthetic planning in organic chemistry, logic synthesis in integrated circuit design, and the classical Sokoban game empirically demonstrate the efficiency of SeeA$^*$, in comparison with the state-of-the-art heuristic search algorithms.