Preference-based reinforcement learning (PbRL) bypasses explicit reward engineering by inferring reward functions from human preference comparisons, enabling better alignment with human intentions. However, humans often struggle to label a clear preference between similar segments, reducing label efficiency and limiting PbRL's real-world applicability. To address this, we propose an offline PbRL method: Contrastive LeArning for ResolvIng Ambiguous Feedback (CLARIFY), which learns a trajectory embedding space that incorporates preference information, ensuring clearly distinguished segments are spaced apart, thus facilitating the selection of more unambiguous queries. Extensive experiments demonstrate that CLARIFY outperforms baselines in both non-ideal teachers and real human feedback settings. Our approach not only selects more distinguished queries but also learns meaningful trajectory embeddings.
Reinforcement learning (RL) has become the dominant paradigm for endowing language models with advanced reasoning capabilities. Despite the substantial empirical gains demonstrated by RL-based training methods like GRPO, a granular understanding of their advantages is still lacking. To address this gap, we introduce a fine-grained analytic framework to dissect the impact of RL on reasoning. Our framework specifically investigates key elements that have been hypothesized to benefit from RL training: (1) plan-following and execution, (2) problem decomposition, and (3) improved reasoning and knowledge utilization. Using this framework, we gain insights beyond mere accuracy. For instance, providing models with explicit step-by-step plans surprisingly degrades performance on the most challenging benchmarks, yet RL-tuned models exhibit greater robustness, experiencing markedly smaller performance drops than their base counterparts. This suggests that RL may not primarily enhance the execution of external plans but rather empower models to formulate and follow internal strategies better suited to their reasoning processes. Conversely, we observe that RL enhances the model's capacity to integrate provided knowledge into its reasoning process, leading to performance improvements across diverse tasks. We also study difficulty, showing improved training by developing new ways to exploit hard problems. Our findings lay a foundation for more principled training and evaluation of reasoning models.
With the proliferation of Internet of Things (IoT) devices, the demand for addressing complex optimization challenges has intensified. The Lyapunov Drift-Plus-Penalty algorithm is a widely adopted approach for ensuring queue stability, and some research has preliminarily explored its integration with reinforcement learning (RL). In this paper, we investigate the adaptation of the Lyapunov Drift-Plus-Penalty algorithm for RL applications, deriving an effective method for combining Lyapunov Drift-Plus-Penalty with RL under a set of common and reasonable conditions through rigorous theoretical analysis. Unlike existing approaches that directly merge the two frameworks, our proposed algorithm, termed Lyapunov drift-plus-penalty method tailored for reinforcement learning with queue stability (LDPTRLQ) algorithm, offers theoretical superiority by effectively balancing the greedy optimization of Lyapunov Drift-Plus-Penalty with the long-term perspective of RL. Simulation results for multiple problems demonstrate that LDPTRLQ outperforms the baseline methods using the Lyapunov drift-plus-penalty method and RL, corroborating the validity of our theoretical derivations. The results also demonstrate that our proposed algorithm outperforms other benchmarks in terms of compatibility and stability.
We propose DRAGO, a novel approach for continual model-based reinforcement learning aimed at improving the incremental development of world models across a sequence of tasks that differ in their reward functions but not the state space or dynamics. DRAGO comprises two key components: Synthetic Experience Rehearsal, which leverages generative models to create synthetic experiences from past tasks, allowing the agent to reinforce previously learned dynamics without storing data, and Regaining Memories Through Exploration, which introduces an intrinsic reward mechanism to guide the agent toward revisiting relevant states from prior tasks. Together, these components enable the agent to maintain a comprehensive and continually developing world model, facilitating more effective learning and adaptation across diverse environments. Empirical evaluations demonstrate that DRAGO is able to preserve knowledge across tasks, achieving superior performance in various continual learning scenarios.
As the era of autonomous agents making decisions on behalf of users unfolds, ensuring contextual integrity (CI) -- what is the appropriate information to share while carrying out a certain task -- becomes a central question to the field. We posit that CI demands a form of reasoning where the agent needs to reason about the context in which it is operating. To test this, we first prompt LLMs to reason explicitly about CI when deciding what information to disclose. We then extend this approach by developing a reinforcement learning (RL) framework that further instills in models the reasoning necessary to achieve CI. Using a synthetic, automatically created, dataset of only $\sim700$ examples but with diverse contexts and information disclosure norms, we show that our method substantially reduces inappropriate information disclosure while maintaining task performance across multiple model sizes and families. Importantly, improvements transfer from this synthetic dataset to established CI benchmarks such as PrivacyLens that has human annotations and evaluates privacy leakage of AI assistants in actions and tool calls.
The discovery of Metal-Organic Frameworks (MOFs) with application-specific properties remains a central challenge in materials chemistry, owing to the immense size and complexity of their structural design space. Conventional computational screening techniques such as molecular simulations and density functional theory (DFT), while accurate, are computationally prohibitive at scale. Machine learning offers an exciting alternative by leveraging data-driven approaches to accelerate materials discovery. The complexity of MOFs, with their extended periodic structures and diverse topologies, creates both opportunities and challenges for generative modeling approaches. To address these challenges, we present a reinforcement learning-enhanced, transformer-based framework for the de novo design of MOFs. Central to our approach is MOFid, a chemically-informed string representation encoding both connectivity and topology, enabling scalable generative modeling. Our pipeline comprises three components: (1) a generative GPT model trained on MOFid sequences, (2) MOFormer, a transformer-based property predictor, and (3) a reinforcement learning (RL) module that optimizes generated candidates via property-guided reward functions. By integrating property feedback into sequence generation, our method drives the model toward synthesizable, topologically valid MOFs with desired functional attributes. This work demonstrates the potential of large language models, when coupled with reinforcement learning, to accelerate inverse design in reticular chemistry and unlock new frontiers in computational MOF discovery.
The control of spatio-temporally chaos is challenging because of high dimensionality and unpredictability. Model-free reinforcement learning (RL) discovers optimal control policies by interacting with the system, typically requiring observations of the full physical state. In practice, sensors often provide only partial and noisy measurements (observations) of the system. The objective of this paper is to develop a framework that enables the control of chaotic systems with partial and noisy observability. The proposed method, data-assimilated model-informed reinforcement learning (DA-MIRL), integrates (i) low-order models to approximate high-dimensional dynamics; (ii) sequential data assimilation to correct the model prediction when observations become available; and (iii) an off-policy actor-critic RL algorithm to adaptively learn an optimal control strategy based on the corrected state estimates. We test DA-MIRL on the spatiotemporally chaotic solutions of the Kuramoto-Sivashinsky equation. We estimate the full state of the environment with (i) a physics-based model, here, a coarse-grained model; and (ii) a data-driven model, here, the control-aware echo state network, which is proposed in this paper. We show that DA-MIRL successfully estimates and suppresses the chaotic dynamics of the environment in real time from partial observations and approximate models. This work opens opportunities for the control of partially observable chaotic systems.
In recent years, deep reinforcement learning (DRL) algorithms have gained traction in home energy management systems. However, their adoption by energy management companies remains limited due to the black-box nature of DRL, which fails to provide transparent decision-making feedback. To address this, explainable reinforcement learning (XRL) techniques have emerged, aiming to make DRL decisions more transparent. Among these, soft differential decision tree (DDT) distillation provides a promising approach due to the clear decision rules they are based on, which can be efficiently computed. However, achieving high performance often requires deep, and completely full, trees, which reduces interpretability. To overcome this, we propose a novel asymmetric soft DDT construction method. Unlike traditional soft DDTs, our approach adaptively constructs trees by expanding nodes only when necessary. This improves the efficient use of decision nodes, which require a predetermined depth to construct full symmetric trees, enhancing both interpretability and performance. We demonstrate the potential of asymmetric DDTs to provide transparent, efficient, and high-performing decision-making in home energy management systems.
This letter proposes an Adversarial Inverse Reinforcement Learning (AIRL)-based energy management method for a smart home, which incorporates an implicit thermal dynamics model. In the proposed method, historical optimal decisions are first generated using a neural network-assisted Hierarchical Model Predictive Control (HMPC) framework. These decisions are then used as expert demonstrations in the AIRL module, which aims to train a discriminator to distinguish expert demonstrations from transitions generated by a reinforcement learning agent policy, while simultaneously updating the agent policy that can produce transitions to confuse the discriminator. The proposed HMPC-AIRL method eliminates the need for explicit thermal dynamics models, prior or predictive knowledge of uncertain parameters, or manually designed reward functions. Simulation results based on real-world traces demonstrate the effectiveness and data efficiency of the proposed method.
Action-dependent individual policies, which incorporate both environmental states and the actions of other agents in decision-making, have emerged as a promising paradigm for achieving global optimality in multi-agent reinforcement learning (MARL). However, the existing literature often adopts auto-regressive action-dependent policies, where each agent's policy depends on the actions of all preceding agents. This formulation incurs substantial computational complexity as the number of agents increases, thereby limiting scalability. In this work, we consider a more generalized class of action-dependent policies, which do not necessarily follow the auto-regressive form. We propose to use the `action dependency graph (ADG)' to model the inter-agent action dependencies. Within the context of MARL problems structured by coordination graphs, we prove that an action-dependent policy with a sparse ADG can achieve global optimality, provided the ADG satisfies specific conditions specified by the coordination graph. Building on this theoretical foundation, we develop a tabular policy iteration algorithm with guaranteed global optimality. Furthermore, we integrate our framework into several SOTA algorithms and conduct experiments in complex environments. The empirical results affirm the robustness and applicability of our approach in more general scenarios, underscoring its potential for broader MARL challenges.
The 3D scene graph models spatial relationships between objects, enabling the agent to efficiently navigate in a partially observable environment and predict the location of the target object.This paper proposes an original framework named SGN-CIRL (3D Scene Graph-Based Reinforcement Learning Navigation) for mapless reinforcement learning-based robot navigation with learnable representation of open-vocabulary 3D scene graph. To accelerate and stabilize the training of reinforcement learning-based algorithms, the framework also employs imitation learning and curriculum learning. The first one enables the agent to learn from demonstrations, while the second one structures the training process by gradually increasing task complexity from simple to more advanced scenarios. Numerical experiments conducted in the Isaac Sim environment showed that using a 3D scene graph for reinforcement learning significantly increased the success rate in difficult navigation cases. The code is open-sourced and available at: https://github.com/Xisonik/Aloha\_graph.
The controller is one of the most important modules in the autonomous driving pipeline, ensuring the vehicle reaches its desired position. In this work, a reinforcement learning based lateral control approach, despite the imperfections in the vehicle models due to measurement errors and simplifications, is presented. Our approach ensures comfortable, efficient, and robust control performance considering the interface between controlling and other modules. The controller consists of the conventional Model Predictive Control (MPC)-PID part as the basis and the demonstrator, and the Deep Reinforcement Learning (DRL) part which leverages the online information from the MPC-PID part. The controller's performance is evaluated in CARLA using the ground truth of the waypoints as inputs. Experimental results demonstrate the effectiveness of the controller when vehicle information is incomplete, and the training of DRL can be stabilized with the demonstration part. These findings highlight the potential to reduce development and integration efforts for autonomous driving pipelines in the future.
Deep reinforcement learning agents are often fragile while humans remain adaptive and flexible to varying scenarios. To bridge this gap, we present EDEN, a biologically inspired navigation framework that integrates learned entorhinal-like grid cell representations and reinforcement learning to enable autonomous navigation. Inspired by the mammalian entorhinal-hippocampal system, EDEN allows agents to perform path integration and vector-based navigation using visual and motion sensor data. At the core of EDEN is a grid cell encoder that transforms egocentric motion into periodic spatial codes, producing low-dimensional, interpretable embeddings of position. To generate these activations from raw sensory input, we combine fiducial marker detections in the lightweight MiniWorld simulator and DINO-based visual features in the high-fidelity Gazebo simulator. These spatial representations serve as input to a policy trained with Proximal Policy Optimization (PPO), enabling dynamic, goal-directed navigation. We evaluate EDEN in both MiniWorld, for rapid prototyping, and Gazebo, which offers realistic physics and perception noise. Compared to baseline agents using raw state inputs (e.g., position, velocity) or standard convolutional image encoders, EDEN achieves a 99% success rate, within the simple scenarios, and >94% within complex floorplans with occluded paths with more efficient and reliable step-wise navigation. In addition, as a replacement of ground truth activations, we present a trainable Grid Cell encoder enabling the development of periodic grid-like patterns from vision and motion sensor data, emulating the development of such patterns within biological mammals. This work represents a step toward biologically grounded spatial intelligence in robotics, bridging neural navigation principles with reinforcement learning for scalable deployment.
Deploying autonomous robots in safety-critical applications requires safety guarantees. Provably safe reinforcement learning is an active field of research which aims to provide such guarantees using safeguards. These safeguards should be integrated during training to prevent a large sim-to-real gap. While there are several approaches for safeguarding sampling-based reinforcement learning, analytic gradient-based reinforcement learning often achieves superior performance and sample efficiency. However, there is no safeguarding approach for this learning paradigm yet. Our work addresses this gap by developing the first effective safeguard for analytic gradient-based reinforcement learning. We analyse existing, differentiable safeguards, adapt them through modified mappings and gradient formulations, and integrate them with a state-of-the-art learning algorithm and a differentiable simulation. We evaluate how different safeguards affect policy optimisation using numerical experiments on two classical control tasks. The results demonstrate safeguarded training without compromising performance.
Hanabi has become a popular game for research when it comes to reinforcement learning (RL) as it is one of the few cooperative card games where you have incomplete knowledge of the entire environment, thus presenting a challenge for a RL agent. We explored different tabular and deep reinforcement learning algorithms to see which had the best performance both against an agent of the same type and also against other types of agents. We establish that certain agents played their highest scoring games against specific agents while others exhibited higher scores on average by adapting to the opposing agent's behavior. We attempted to quantify the conditions under which each algorithm provides the best advantage and identified the most interesting interactions between agents of different types. In the end, we found that temporal difference (TD) algorithms had better overall performance and balancing of play types compared to tabular agents. Specifically, tabular Expected SARSA and deep Q-Learning agents showed the best performance.
Large Language Models (LLMs) increasingly rely on Supervised Fine-Tuning (SFT) and Reinforcement Learning from Human Feedback (RLHF) to align model responses with human preferences. While RLHF employs a reinforcement learning approach with a separate reward model, SFT uses human-curated datasets for supervised learning. Both approaches traditionally depend on small, vetted groups of annotators, making them costly, prone to bias, and limited in scalability. We propose an open, crowd-sourced fine-tuning framework that addresses these limitations by enabling broader feedback collection for SFT without extensive annotator training. Our framework promotes incentive fairness via a point-based reward system correlated with Shapley values and guides model convergence through iterative model updates. Our multi-model selection framework demonstrates up to a 55% reduction in target distance over single-model selection, enabling subsequent experiments that validate our point-based reward mechanism's close alignment with Shapley values (a well-established method for attributing individual contributions) thereby supporting fair and scalable participation.
We propose a novel composite reward function for reinforcement learning in financial trading that balances return and risk using four differentiable terms: annualized return downside risk differential return and the Treynor ratio Unlike single metric objectives for example the Sharpe ratio our formulation is modular and parameterized by weights w1 w2 w3 and w4 enabling practitioners to encode diverse investor preferences We tune these weights via grid search to target specific risk return profiles We derive closed form gradients for each term to facilitate gradient based training and analyze key theoretical properties including monotonicity boundedness and modularity This framework offers a general blueprint for building robust multi objective reward functions in complex trading environments and can be extended with additional risk measures or adaptive weighting
Current Large Language Models (LLMs) exhibit significant limitations, notably in structured, interpretable, and verifiable medical reasoning, alongside practical deployment challenges related to computational resources and data privacy. This report focused on the development of WiNGPT-3.0, the 32-billion parameter LLMs, engineered with the objective of enhancing its capacity for medical reasoning and exploring its potential for effective integration within healthcare IT infrastructures. The broader aim is to advance towards clinically applicable models. The approach involved a multi-stage training pipeline tailored for general, medical, and clinical reasoning. This pipeline incorporated supervised fine-tuning (SFT) and reinforcement learning (RL), leveraging curated Long Chain-of-Thought (CoT) datasets, auxiliary reward models, and an evidence-based diagnostic chain simulation. WiNGPT-3.0 demonstrated strong performance: specific model variants achieved scores of 66.6 on MedCalc and 87.1 on MedQA-USMLE. Furthermore, targeted training improved performance on a clinical reasoning task from a baseline score of 58.1 to 62.5. These findings suggest that reinforcement learning, even when applied with a limited dataset of only a few thousand examples, can enhance medical reasoning accuracy. Crucially, this demonstration of RL's efficacy with limited data and computation paves the way for more trustworthy and practically deployable LLMs within clinical workflows and health information infrastructures.
Emerging embodied AI applications, such as wearable cameras and autonomous agents, have underscored the need for robust reasoning from first person video streams. We introduce EgoVLM, a vision-language model specifically designed to integrate visual comprehension and spatial-temporal reasoning within egocentric video contexts. EgoVLM is fine-tuned via Group Relative Policy Optimization (GRPO), a reinforcement learning method adapted to align model outputs with human-like reasoning steps. Following DeepSeek R1-Zero's approach, we directly tune using RL without any supervised fine-tuning phase on chain-of-thought (CoT) data. We evaluate EgoVLM on egocentric video question answering benchmarks and show that domain-specific training substantially improves performance over general-purpose VLMs. Our EgoVLM-3B, trained exclusively on non-CoT egocentric data, outperforms the base Qwen2.5-VL 3B and 7B models by 14.33 and 13.87 accuracy points on the EgoSchema benchmark, respectively. By explicitly generating reasoning traces, EgoVLM enhances interpretability, making it well-suited for downstream applications. Furthermore, we introduce a novel keyframe-based reward that incorporates salient frame selection to guide reinforcement learning optimization. This reward formulation opens a promising avenue for future exploration in temporally grounded egocentric reasoning.
Table reasoning, encompassing tasks such as table question answering, fact verification, and text-to-SQL, requires precise understanding of structured tabular data, coupled with numerical computation and code manipulation for effective inference. Supervised fine-tuning (SFT) approaches have achieved notable success but often struggle with generalization and robustness due to biases inherent in imitative learning. We introduce Reasoning-Table, the first application of reinforcement learning (RL) to table reasoning, achieving state-of-the-art performance. Through rigorous data preprocessing, reward design, and tailored training strategies, our method leverages simple rule-based outcome rewards to outperform SFT across multiple benchmarks. Unified training across diverse tasks enables Reasoning-Table to emerge as a robust table reasoning large language model, surpassing larger proprietary models like Claude-3.7-Sonnet by 4.0% on table reasoning benchmarks. The approach also achieves excellent performance on text-to-SQL tasks, reaching 68.3% performance on the BIRD dev dataset with a 7B model. Further experiments demonstrate that Reasoning-Table enhances the model's generalization capabilities and robustness.