Q-learning is widely recognized as an effective approach for synthesizing controllers to achieve specific goals. However, handling challenges posed by continuous state-action spaces remains an ongoing research focus. This paper presents a systematic analysis that highlights a major drawback in space discretization methods. To address this challenge, the paper proposes a symbolic model that represents behavioral relations, such as alternating simulation from abstraction to the controlled system. This relation allows for seamless application of the synthesized controller based on abstraction to the original system. Introducing a novel Q-learning technique for symbolic models, the algorithm yields two Q-tables encoding optimal policies. Theoretical analysis demonstrates that these Q-tables serve as both upper and lower bounds on the Q-values of the original system with continuous spaces. Additionally, the paper explores the correlation between the parameters of the space abstraction and the loss in Q-values. The resulting algorithm facilitates achieving optimality within an arbitrary accuracy, providing control over the trade-off between accuracy and computational complexity. The obtained results provide valuable insights for selecting appropriate learning parameters and refining the controller. The engineering relevance of the proposed Q-learning based symbolic model is illustrated through two case studies.

Causal dynamics learning has recently emerged as a promising approach to enhancing robustness in reinforcement learning (RL). Typically, the goal is to build a dynamics model that makes predictions based on the causal relationships among the entities. Despite the fact that causal connections often manifest only under certain contexts, existing approaches overlook such fine-grained relationships and lack a detailed understanding of the dynamics. In this work, we propose a novel dynamics model that infers fine-grained causal structures and employs them for prediction, leading to improved robustness in RL. The key idea is to jointly learn the dynamics model with a discrete latent variable that quantizes the state-action space into subgroups. This leads to recognizing meaningful context that displays sparse dependencies, where causal structures are learned for each subgroup throughout the training. Experimental results demonstrate the robustness of our method to unseen states and locally spurious correlations in downstream tasks where fine-grained causal reasoning is crucial. We further illustrate the effectiveness of our subgroup-based approach with quantization in discovering fine-grained causal relationships compared to prior methods.

Efficiently tackling combinatorial reasoning problems, particularly the notorious NP-hard tasks, remains a significant challenge for AI research. Recent efforts have sought to enhance planning by incorporating hierarchical high-level search strategies, known as subgoal methods. While promising, their performance against traditional low-level planners is inconsistent, raising questions about their application contexts. In this study, we conduct an in-depth exploration of subgoal-planning methods for combinatorial reasoning. We identify the attributes pivotal for leveraging the advantages of high-level search: hard-to-learn value functions, complex action spaces, presence of dead ends in the environment, or using data collected from diverse experts. We propose a consistent evaluation methodology to achieve meaningful comparisons between methods and reevaluate the state-of-the-art algorithms.

We consider the problem of multi-objective alignment of foundation models with human preferences, which is a critical step towards helpful and harmless AI systems. However, it is generally costly and unstable to fine-tune large foundation models using reinforcement learning (RL), and the multi-dimensionality, heterogeneity, and conflicting nature of human preferences further complicate the alignment process. In this paper, we introduce Rewards-in-Context (RiC), which conditions the response of a foundation model on multiple rewards in its prompt context and applies supervised fine-tuning for alignment. The salient features of RiC are simplicity and adaptivity, as it only requires supervised fine-tuning of a single foundation model and supports dynamic adjustment for user preferences during inference time. Inspired by the analytical solution of an abstracted convex optimization problem, our dynamic inference-time adjustment method approaches the Pareto-optimal solution for multiple objectives. Empirical evidence demonstrates the efficacy of our method in aligning both Large Language Models (LLMs) and diffusion models to accommodate diverse rewards with only around 10% GPU hours compared with multi-objective RL baseline.

On-policy reinforcement learning methods, like Trust Region Policy Optimization (TRPO) and Proximal Policy Optimization (PPO), often demand extensive data per update, leading to sample inefficiency. This paper introduces Reflective Policy Optimization (RPO), a novel on-policy extension that amalgamates past and future state-action information for policy optimization. This approach empowers the agent for introspection, allowing modifications to its actions within the current state. Theoretical analysis confirms that policy performance is monotonically improved and contracts the solution space, consequently expediting the convergence procedure. Empirical results demonstrate RPO's feasibility and efficacy in two reinforcement learning benchmarks, culminating in superior sample efficiency. The source code of this work is available at https://github.com/Edgargan/RPO.

Learning a good representation is a crucial challenge for Reinforcement Learning (RL) agents. Self-predictive learning provides means to jointly learn a latent representation and dynamics model by bootstrapping from future latent representations (BYOL). Recent work has developed theoretical insights into these algorithms by studying a continuous-time ODE model for self-predictive representation learning under the simplifying assumption that the algorithm depends on a fixed policy (BYOL-$\Pi$); this assumption is at odds with practical instantiations of such algorithms, which explicitly condition their predictions on future actions. In this work, we take a step towards bridging the gap between theory and practice by analyzing an action-conditional self-predictive objective (BYOL-AC) using the ODE framework, characterizing its convergence properties and highlighting important distinctions between the limiting solutions of the BYOL-$\Pi$ and BYOL-AC dynamics. We show how the two representations are related by a variance equation. This connection leads to a novel variance-like action-conditional objective (BYOL-VAR) and its corresponding ODE. We unify the study of all three objectives through two complementary lenses; a model-based perspective, where each objective is shown to be equivalent to a low-rank approximation of certain dynamics, and a model-free perspective, which establishes relationships between the objectives and their respective value, Q-value, and advantage function. Our empirical investigations, encompassing both linear function approximation and Deep RL environments, demonstrates that BYOL-AC is better overall in a variety of different settings.

Production scheduling is an essential task in manufacturing, with Reinforcement Learning (RL) emerging as a key solution. In a previous work, RL was utilized to solve an extended permutation flow shop scheduling problem (PFSSP) for a real-world production line with two stages, linked by a central buffer. The RL agent was trained to sequence equallysized product batches to minimize setup efforts and idle times. However, the substantial impact caused by varying the size of these product batches has not yet been explored. In this follow-up study, we investigate the effects of varying batch sizes, exploring both the quality of solutions and the training dynamics of the RL agent. The results demonstrate that it is possible to methodically identify reasonable boundaries for the batch size. These boundaries are determined on one side by the increasing sample complexity associated with smaller batch sizes, and on the other side by the decreasing flexibility of the agent when dealing with larger batch sizes. This provides the practitioner the ability to make an informed decision regarding the selection of an appropriate batch size. Moreover, we introduce and investigate two new curriculum learning strategies to enable the training with small batch sizes. The findings of this work offer the potential for application in several industrial use cases with comparable scheduling problems.

Learning representations for reinforcement learning (RL) has shown much promise for continuous control. We propose an efficient representation learning method using only a self-supervised latent-state consistency loss. Our approach employs an encoder and a dynamics model to map observations to latent states and predict future latent states, respectively. We achieve high performance and prevent representation collapse by quantizing the latent representation such that the rank of the representation is empirically preserved. Our method, named iQRL: implicitly Quantized Reinforcement Learning, is straightforward, compatible with any model-free RL algorithm, and demonstrates excellent performance by outperforming other recently proposed representation learning methods in continuous control benchmarks from DeepMind Control Suite.

Deep Reinforcement Learning (DRL) has emerged as a promising approach to enhancing motion control and decision-making through a wide range of robotic applications. While prior research has demonstrated the efficacy of DRL algorithms in facilitating autonomous mapless navigation for aerial and terrestrial mobile robots, these methods often grapple with poor generalization when faced with unknown tasks and environments. This paper explores the impact of the Delayed Policy Updates (DPU) technique on fostering generalization to new situations, and bolstering the overall performance of agents. Our analysis of DPU in aerial and terrestrial mobile robots reveals that this technique significantly curtails the lack of generalization and accelerates the learning process for agents, enhancing their efficiency across diverse tasks and unknown scenarios.

We propose a discrete time graphon game formulation on continuous state and action spaces using a representative player to study stochastic games with heterogeneous interaction among agents. This formulation admits both philosophical and mathematical advantages, compared to a widely adopted formulation using a continuum of players. We prove the existence and uniqueness of the graphon equilibrium with mild assumptions, and show that this equilibrium can be used to construct an approximate solution for finite player game on networks, which is challenging to analyze and solve due to curse of dimensionality. An online oracle-free learning algorithm is developed to solve the equilibrium numerically, and sample complexity analysis is provided for its convergence.