In this paper, we study the finite-sample statistical rates of distributional temporal difference (TD) learning with linear function approximation. The purpose of distributional TD learning is to estimate the return distribution of a discounted Markov decision process for a given policy. Previous works on statistical analysis of distributional TD learning focus mainly on the tabular case. We first consider the linear function approximation setting and conduct a fine-grained analysis of the linear-categorical Bellman equation. Building on this analysis, we further incorporate variance reduction techniques in our new algorithms to establish tight sample complexity bounds independent of the support size $K$ when $K$ is large. Our theoretical results imply that, when employing distributional TD learning with linear function approximation, learning the full distribution of the return function from streaming data is no more difficult than learning its expectation. This work provide new insights into the statistical efficiency of distributional reinforcement learning algorithms.

In this study, we tackle the challenging task of predicting secondary structures from protein primary sequences, a pivotal initial stride towards predicting tertiary structures, while yielding crucial insights into protein activity, relationships, and functions. Existing methods often utilize extensive sets of unlabeled amino acid sequences. However, these approaches neither explicitly capture nor harness the accessible protein 3D structural data, which is recognized as a decisive factor in dictating protein functions. To address this, we utilize protein residue graphs and introduce various forms of sequential or structural connections to capture enhanced spatial information. We adeptly combine Graph Neural Networks (GNNs) and Language Models (LMs), specifically utilizing a pre-trained transformer-based protein language model to encode amino acid sequences and employing message-passing mechanisms like GCN and R-GCN to capture geometric characteristics of protein structures. Employing convolution within a specific node's nearby region, including relations, we stack multiple con-volutional layers to efficiently learn combined insights from the protein's spatial graph, revealing intricate interconnections and dependencies in its structural To assess our model's performance, we employed the training dataset provided by NetSurfP-2.0, which outlines secondary structure in 3-and 8-states. Extensive experiments show that our proposed model, SSRGNet surpasses the baseline on f1-scores. Introduction Proteins serve as essential components within cells and are involved in various applications, spanning from therapeutics to materials. They are composed of a sequence of amino acids that fold into distinct shapes. With the development of affordable sequencing technologies [1, 2], a substantial number of novel protein sequences have been identified in recent times. However, annotating the functional properties of a newly discovered protein sequence is still a laborious and expensive process. Thus, there is a need for reliable and efficient computational methods to accurately predict and assign functions to proteins, thereby bridging the gap between sequence information and functional knowledge. The analysis of protein structure, particularly the tertiary structure, is highly significant for practical applications related to proteins, such as understanding their functions and designing drugs [3].

The problem of cooperative data censoring in battery-powered multihop sensor networks is analyzed in this paper. We are interested in scenarios where nodes generate messages (which are related to the sensor measurements) that can be graded with some importance value. Less important messages can be censored in order to save energy for later communications. The problem is modeled using a joint Markov Decision Process of the whole network dynamics, and a theoretically optimal censoring policy, which maximizes a long-term reward, is found. Though the optimal censoring rules are computationally prohibitive, our analysis suggests that, under some conditions, they can be approximated by a finite collection of constant-threshold rules. A centralized algorithm for the computation of these thresholds is proposed. The experimental simulations show that cooperative censoring policies are energy-efficient, and outperform other non-cooperative schemes.

Accurate healthcare prediction is critical for improving patient outcomes and reducing operational costs. Bolstered by growing reasoning capabilities, large language models (LLMs) offer a promising path to enhance healthcare predictions by drawing on their rich parametric knowledge. However, LLMs are prone to factual inaccuracies due to limitations in the reliability and coverage of their embedded knowledge. While retrieval-augmented generation (RAG) frameworks, such as GraphRAG and its variants, have been proposed to mitigate these issues by incorporating external knowledge, they face two key challenges in the healthcare scenario: (1) identifying the clinical necessity to activate the retrieval mechanism, and (2) achieving synergy between the retriever and the generator to craft contextually appropriate retrievals. To address these challenges, we propose GHAR, a \underline{g}enerative \underline{h}ierarchical \underline{a}gentic \underline{R}AG framework that simultaneously resolves when to retrieve and how to optimize the collaboration between submodules in healthcare. Specifically, for the first challenge, we design a dual-agent architecture comprising Agent-Top and Agent-Low. Agent-Top acts as the primary physician, iteratively deciding whether to rely on parametric knowledge or to initiate retrieval, while Agent-Low acts as the consulting service, summarising all task-relevant knowledge once retrieval was triggered. To tackle the second challenge, we innovatively unify the optimization of both agents within a formal Markov Decision Process, designing diverse rewards to align their shared goal of accurate prediction while preserving their distinct roles. Extensive experiments on three benchmark datasets across three popular tasks demonstrate our superiority over state-of-the-art baselines, highlighting the potential of hierarchical agentic RAG in advancing healthcare systems.

Task decomposition has shown promise in complex cooperative multi-agent reinforcement learning (MARL) tasks, which enables efficient hierarchical learning for long-horizon tasks in dynamic and uncertain environments. However, learning dynamic task decomposition from scratch generally requires a large number of training samples, especially exploring the large joint action space under partial observability. In this paper, we present the Conditional Diffusion Model for Dynamic Task Decomposition (C$\text{D}^\text{3}$T), a novel two-level hierarchical MARL framework designed to automatically infer subtask and coordination patterns. The high-level policy learns subtask representation to generate a subtask selection strategy based on subtask effects. To capture the effects of subtasks on the environment, C$\text{D}^\text{3}$T predicts the next observation and reward using a conditional diffusion model. At the low level, agents collaboratively learn and share specialized skills within their assigned subtasks. Moreover, the learned subtask representation is also used as additional semantic information in a multi-head attention mixing network to enhance value decomposition and provide an efficient reasoning bridge between individual and joint value functions. Experimental results on various benchmarks demonstrate that C$\text{D}^\text{3}$T achieves better performance than existing baselines.

Multi-agent reinforcement learning (MARL) has shown promise for large-scale network control, yet existing methods face two major limitations. First, they typically rely on assumptions leading to decay properties of local agent interactions, limiting their ability to capture long-range dependencies such as cascading power failures or epidemic outbreaks. Second, most approaches lack generalizability across network topologies, requiring retraining when applied to new graphs. We introduce STACCA (Shared Transformer Actor-Critic with Counterfactual Advantage), a unified transformer-based MARL framework that addresses both challenges. STACCA employs a centralized Graph Transformer Critic to model long-range dependencies and provide system-level feedback, while its shared Graph Transformer Actor learns a generalizable policy capable of adapting across diverse network structures. Further, to improve credit assignment during training, STACCA integrates a novel counterfactual advantage estimator that is compatible with state-value critic estimates. We evaluate STACCA on epidemic containment and rumor-spreading network control tasks, demonstrating improved performance, network generalization, and scalability. These results highlight the potential of transformer-based MARL architectures to achieve scalable and generalizable control in large-scale networked systems.

This paper investigates a project with stochastic activity durations and cash flows under discrete scenarios, where activities must satisfy precedence constraints generating cash inflows and outflows. The objective is to maximize expected net present value (NPV) by accelerating inflows and deferring outflows. We formulate the problem as a discrete-time Markov Decision Process (MDP) and propose a Double Deep Q-Network (DDQN) approach. Comparative experiments demonstrate that DDQN outperforms traditional rigid and dynamic strategies, particularly in large-scale or highly uncertain environments, exhibiting superior computational capability, policy reliability, and adaptability. Ablation studies further reveal that the dual-network architecture mitigates overestimation of action values, while the target network substantially improves training convergence and robustness. These results indicate that DDQN not only achieves higher expected NPV in complex project optimization but also provides a reliable framework for stable and effective policy implementation.

This work investigates resource optimization in heterogeneous satellite clusters performing autonomous Earth Observation (EO) missions using Reinforcement Learning (RL). In the proposed setting, two optical satellites and one Synthetic Aperture Radar (SAR) satellite operate cooperatively in low Earth orbit to capture ground targets and manage their limited onboard resources efficiently. Traditional optimization methods struggle to handle the real-time, uncertain, and decentralized nature of EO operations, motivating the use of RL and Multi-Agent Reinforcement Learning (MARL) for adaptive decision-making. This study systematically formulates the optimization problem from single-satellite to multi-satellite scenarios, addressing key challenges including energy and memory constraints, partial observability, and agent heterogeneity arising from diverse payload capabilities. Using a near-realistic simulation environment built on the Basilisk and BSK-RL frameworks, we evaluate the performance and stability of state-of-the-art MARL algorithms such as MAPPO, HAPPO, and HATRPO. Results show that MARL enables effective coordination across heterogeneous satellites, balancing imaging performance and resource utilization while mitigating non-stationarity and inter-agent reward coupling. The findings provide practical insights into scalable, autonomous satellite operations and contribute a foundation for future research on intelligent EO mission planning under heterogeneous and dynamic conditions.

Are LLMs The W ay Forward? Abstract--Autonomous vehicle navigation in complex environments such as dense and fast-moving highways and merging scenarios remains an active area of research. In the past decade, many planning and control approaches have used reinforcement learning (RL) with notable success. However, a key limitation of RL is its reliance on well-specified reward functions, which often fail to capture the full semantic and social complexity of diverse, out-of-distribution situations. As a result, a rapidly growing line of research explores using Large Language Models (LLMs) to replace or supplement RL for direct planning and control, on account of their ability to reason about rich semantic context. However, LLMs present significant drawbacks: they can be unstable in zero-shot safety-critical settings, produce inconsistent outputs, and often depend on expensive API calls with network latency. This motivates our investigation into whether small, locally deployed LLMs ( 14B parameters) can meaningfully support autonomous highway driving through reward shaping rather than direct control. These models are attractive for practical deployment as they can run on a single GPU and avoid external API dependencies. We present a case study comparing RL-only, LLM-only, and hybrid approaches, where LLMs augment RL rewards by scoring state-action transitions during training, while standard RL policies execute at test time.

We define the problem of linear Contextual Stochastic Shortest Path (CSSP), where at the beginning of each episode, the learner observes an adversarially chosen context that determines the MDP through a fixed but unknown linear function. The learner's objective is to reach a designated goal state with minimal expected cumulative loss, despite having no prior knowledge of the transition dynamics, loss functions, or the mapping from context to MDP. In this work, we propose LR-CSSP, an algorithm that achieves a regret bound of $\widetilde{O}(K^{2/3} d^{2/3} |S| |A|^{1/3} B_\star^2 T_\star \log (1/ δ))$, where $K$ is the number of episodes, $d$ is the context dimension, $S$ and $A$ are the sets of states and actions respectively, $B_\star$ bounds the optimal cumulative loss and $T_\star$, unknown to the learner, bounds the expected time for the optimal policy to reach the goal. In the case where all costs exceed $\ell_{\min}$, LR-CSSP attains a regret of $\widetilde O(\sqrt{K \cdot d^2 |S|^3 |A| B_\star^3 \log(1/δ)/\ell_{\min}})$. Unlike in contextual finite-horizon MDPs, where limited knowledge primarily leads to higher losses and regret, in the CSSP setting, insufficient knowledge can also prolong episodes and may even lead to non-terminating episodes. Our analysis reveals that LR-CSSP effectively handles continuous context spaces, while ensuring all episodes terminate within a reasonable number of time steps.