We introduce the Continuous Arcade Learning Environment (CALE), an extension of the well-known Arcade Learning Environment (ALE) [Bellemare et al., 2013]. The CALE uses the same underlying emulator of the Atari 2600 gaming system (Stella), but adds support for continuous actions. This enables the benchmarking and evaluation of continuous-control agents (such as PPO [Schulman et al., 2017] and SAC [Haarnoja et al., 2018]) and value-based agents (such as DQN [Mnih et al., 2015] and Rainbow [Hessel et al., 2018]) on the same environment suite. We provide a series of open questions and research directions that CALE enables, as well as initial baseline results using Soft Actor-Critic.

Offline-to-online reinforcement learning (RL) leverages both pre-trained offline policies and online policies trained for downstream tasks, aiming to improve data efficiency and accelerate performance enhancement. An existing approach, Policy Expansion (PEX), utilizes a policy set composed of both policies without modifying the offline policy for exploration and learning. However, this approach fails to ensure sufficient learning of the online policy due to an excessive focus on exploration with both policies. Since the pre-trained offline policy can assist the online policy in exploiting a downstream task based on its prior experience, it should be executed effectively and tailored to the specific requirements of the downstream task. In contrast, the online policy, with its immature behavioral strategy, has the potential for exploration during the training phase. Therefore, our research focuses on harmonizing the advantages of the offline policy, termed exploitation, with those of the online policy, referred to as exploration, without modifying the offline policy. In this study, we propose an innovative offline-to-online RL method that employs a non-monolithic exploration approach. Our methodology demonstrates superior performance compared to PEX. The code for this comparison is readily available.

Many imitation learning (IL) algorithms use inverse reinforcement learning (IRL) to infer a reward function that aligns with the demonstrations. However, the inferred reward function often fails to capture the underlying task objective. In this paper, we propose a novel framework for IRL-based IL that prioritizes task alignment over conventional data alignment. Our framework is a semi-supervised approach that leverages expert demonstrations as weak supervision signals to derive a set of candidate reward functions that align with the task rather than only with the data. It adopts an adversarial mechanism to train a policy with this set of reward functions to gain a collective validation of the policy's ability to accomplish the task. We provide theoretical insights into this framework's ability to mitigate task-reward misalignment and present a practical implementation. Our experimental results show that our framework outperforms conventional IL baselines in complex and transfer learning scenarios. The complete code are available at https://github.com/zwc662/PAGAR.

We introduce anytime constraints to the multi-agent setting with the corresponding solution concept being anytime-constrained equilibrium (ACE). Then, we present a comprehensive theory of anytime-constrained Markov games, which includes (1) a computational characterization of feasible policies, (2) a fixed-parameter tractable algorithm for computing ACE, and (3) a polynomial-time algorithm for approximately computing feasible ACE. Since computing a feasible policy is NP-hard even for two-player zero-sum games, our approximation guarantees are the best possible under worst-case analysis. We also develop the first theory of efficient computation for action-constrained Markov games, which may be of independent interest.

Emerging research in Pluralistic Artificial Intelligence (AI) alignment seeks to address how intelligent systems can be designed and deployed in accordance with diverse human needs and values. We contribute to this pursuit with a dynamic approach for aligning AI with diverse and shifting user preferences through Multi-Objective Reinforcement Learning (MORL), via post-learning policy selection adjustment. In this paper, we introduce the proposed framework for this approach, outline its anticipated advantages and assumptions, and discuss technical details about the implementation. We also examine the broader implications of adopting a retroactive alignment approach through the sociotechnical systems perspective.

The application of reinforcement learning in traffic signal control (TSC) has been extensively researched and yielded notable achievements. However, most existing works for TSC assume that traffic data from all surrounding intersections is fully and continuously available through sensors. In real-world applications, this assumption often fails due to sensor malfunctions or data loss, making TSC with missing data a critical challenge. To meet the needs of practical applications, we introduce DiffLight, a novel conditional diffusion model for TSC under data-missing scenarios in the offline setting. Specifically, we integrate two essential sub-tasks, i.e., traffic data imputation and decision-making, by leveraging a Partial Rewards Conditioned Diffusion (PRCD) model to prevent missing rewards from interfering with the learning process. Meanwhile, to effectively capture the spatial-temporal dependencies among intersections, we design a Spatial-Temporal transFormer (STFormer) architecture. In addition, we propose a Diffusion Communication Mechanism (DCM) to promote better communication and control performance under data-missing scenarios. Extensive experiments on five datasets with various data-missing scenarios demonstrate that DiffLight is an effective controller to address TSC with missing data. The code of DiffLight is released at https://github.com/lokol5579/DiffLight-release.

We develop a generic policy gradient method with the global optimality guarantee for robust Markov Decision Processes (MDPs). While policy gradient methods are widely used for solving dynamic decision problems due to their scalable and efficient nature, adapting these methods to account for model ambiguity has been challenging, often making it impractical to learn robust policies. This paper introduces a novel policy gradient method, Double-Loop Robust Policy Mirror Descent (DRPMD), for solving robust MDPs. DRPMD employs a general mirror descent update rule for the policy optimization with adaptive tolerance per iteration, guaranteeing convergence to a globally optimal policy. We provide a comprehensive analysis of DRPMD, including new convergence results under both direct and softmax parameterizations, and provide novel insights into the inner problem solution through Transition Mirror Ascent (TMA). Additionally, we propose innovative parametric transition kernels for both discrete and continuous state-action spaces, broadening the applicability of our approach.

In this work, we prove that, in linear MDPs, the feature dimension $d$ is lower bounded by $S/U$ in order to aptly represent transition probabilities, where $S$ is the size of the state space and $U$ is the maximum size of directly reachable states. Hence, $d$ can still scale with $S$ depending on the direct reachability of the environment. To address this limitation of linear MDPs, we propose a novel structural aggregation framework based on dynamics, named as the "dynamics aggregation". For this newly proposed framework, we design a provably efficient hierarchical reinforcement learning algorithm in linear function approximation that leverages aggregated sub-structures. Our proposed algorithm exhibits statistical efficiency, achieving a regret of $ \tilde{O} ( d_{\psi}^{3/2} H^{3/2}\sqrt{ N T} )$, where $d_{\psi}$ represents the feature dimension of aggregated subMDPs and $N$ signifies the number of aggregated subMDPs. We establish that the condition $d_{\psi}^3 N \ll d^{3}$ is readily met in most real-world environments with hierarchical structures, enabling a substantial improvement in the regret bound compared to LSVI-UCB, which enjoys a regret of $ \tilde{O} (d^{3/2} H^{3/2} \sqrt{ T})$. To the best of our knowledge, this work presents the first HRL algorithm with linear function approximation that offers provable guarantees.

Goal-conditioned reinforcement learning is a powerful way to control an AI agent's behavior at runtime. That said, popular goal representations, e.g., target states or natural language, are either limited to Markovian tasks or rely on ambiguous task semantics. We propose representing temporal goals using compositions of deterministic finite automata (cDFAs) and use cDFAs to guide RL agents. cDFAs balance the need for formal temporal semantics with ease of interpretation: if one can understand a flow chart, one can understand a cDFA. On the other hand, cDFAs form a countably infinite concept class with Boolean semantics, and subtle changes to the automaton can result in very different tasks, making them difficult to condition agent behavior on. To address this, we observe that all paths through a DFA correspond to a series of reach-avoid tasks and propose pre-training graph neural network embeddings on "reach-avoid derived" DFAs. Through empirical evaluation, we demonstrate that the proposed pre-training method enables zero-shot generalization to various cDFA task classes and accelerated policy specialization without the myopic suboptimality of hierarchical methods.

This paper explores the mechanistic interpretability of reinforcement learning (RL) agents through an analysis of a neural network trained on procedural maze environments. By dissecting the network's inner workings, we identified fundamental features like maze walls and pathways, forming the basis of the model's decision-making process. A significant observation was the goal misgeneralization, where the RL agent developed biases towards certain navigation strategies, such as consistently moving towards the top right corner, even in the absence of explicit goals. Using techniques like saliency mapping and feature mapping, we visualized these biases. We furthered this exploration with the development of novel tools for interactively exploring layer activations.