Offline Reinforcement Learning (RL) has emerged as a promising paradigm to overcome the limitations of conventional RL methods that depend on extensive and often unsafe online interactions with the environment. By learning policies exclusively from pre-collected datasets, Offline RL aligns with the growing emphasis on data efficiency and safety in machine learning. However, existing off-policy RL algorithms encounter substantial difficulties when trained purely on offline data, primarily due to distributional shift between the training dataset and the learned policy. This issue becomes more pronounced with high-dimensional function approximation, leading to degraded performance and poor generalization. This survey provides a comprehensive overview of recent advances in Offline RL, with a particular focus on the challenges of distribution shift, generalization, and out-of-distribution (OOD) actions. We review the evolution of this field, discuss data limitations, and analyze the historical and theoretical foundations of distributional shift in Offline RL. Furthermore, we categorize existing approaches into four major groups including Q-value restriction methods, uncertainty-based Q-restriction methods, policy constraint methods, and uncertainty-based policy constraint methods, highlighting their core principles and practical implications. We also summarize common benchmarks such as D4RL, discuss dataset enhancement strategies, and examine evaluation metrics and computational efficiency. Through this analysis, the survey offers a unified perspective on current solutions and identifies open challenges, providing valuable guidance for advancing robust, generalizable, and trustworthy Offline RL in domains such as robotics, healthcare, and autonomous systems.

Misclassifications in spam and phishing detection are very harmful, as false negatives expose users to attacks while false positives degrade trust. Existing uncertainty-based detectors can flag potential errors, but may be deceived and offer limited interpretability. This paper presents X-MAP, an eXplainable Misclassification Analysis and Profilling framework that reveals topic-level semantic patterns behind model failures. X-MAP combines SHAP-based feature attributions with non-negative matrix factorization to build interpretable topic profiles for reliably classified spam/phishing and legitimate messages, and measures each message’s deviation from these profiles using Jensen–Shannon divergence. Experiments on SMS and phishing datasets show that misclassified messages exhibit at least twice the divergence of correctly classified ones. As a detector, X-MAP achieves up to 0.98 AUROC and lowers the false-rejection rate at 95

Traffic flow prediction is a core task in intelligent transportation systems. However, existing methods still face limitations in addressing complex spatiotemporal dependencies and improving computational efficiency. To tackle these issues, this study proposes an adaptive hypergraph framework based on feature sharing. The framework incorporates multi-scale temporal convolution layers, which can capture dynamic features at different time granularities. Additionally, by introducing adaptive hypergraph convolution, the model achieves more accurate modeling of complex spatial dependencies, generating relatively stable spatial structure representations. To further enhance the model’s flexibility, we propose a hypergraph masking mechanism, enabling the hypergraph to dynamically adapt to changes in spatial topology. Moreover, this study innovatively introduces a feature-sharing strategy, reusing previously extracted features during periods of relatively stable traffic patterns, thereby achieving feature reuse along the temporal dimension. This approach not only avoids redundant computations, significantly reducing computational costs, but also enhances the model’s robustness and generalization ability. Experimental results demonstrate that the proposed framework achieves higher prediction accuracy while substantially reducing computation time across three benchmark datasets, showcasing excellent time efficiency.

Uncertainty quantification (UQ) is critical, since LLMs frequently produce confident-sounding but incorrect responses even in high-stakes applications. Existing UQ methods for LLMs, such as Semantic Entropy and SelfCheckGPT, rely on multiple sampled generations, resulting in high computational cost and limited scalability. Recent work suggests that LLM confidence is closely tied to prompt sensitivity: confident predictions remain stable under minor instruction changes, whereas uncertain ones exhibit greater variability. Building on this insight, we introduce C^3 , a fast contrastive confidence-based UQ method that estimates the uncertainty of an LLM’s response to a user prompt, given a system instruction, by measuring how its token-level output distributions change when that system instruction is perturbed by a learned instruction-level modification. This learned system-instruction perturbation is optimized to maximize the token-level distributional contrast between confident and hallucination-prone regions within the same generated response, and it generalizes across user inputs, models, and tasks. With only one additional forward pass, C^3 achieves comparable or superior hallucination-detection performance to state-of-the-art baselines while reducing inference cost by over 97 C^3 as an efficient and effective tool for uncertainty estimation in modern LLMs.

Construction Engineering and Management (CEM) relies heavily on large volumes of regulatory standards, yet the lack of systematic evaluation frameworks limits the reliable deployment of large language models (LLMs) for automated regulatory understanding and decision support. To advance automation in construction knowledge processing, this paper presents a comprehensive benchmark for evaluating large language models in construction engineering and management, termed CemEval. CemEval is constructed from 573 valid national and industry standards, covering design, construction, and management regulations. This paper introduces a Clause-to-Task paradigm that automatically transforms regulatory clauses into four executable task types: Multiple-Choice, True or False, Fill-in-the-Blank, and Short Answer tasks. These tasks are further categorized into numerical and non-numerical subsets, enabling systematic assessment of automated capabilities in numerical computation, unit conversion, regulatory reasoning, and semantic understanding. The resulting benchmark comprises 34,537 high-quality task samples, establishing a multi-task, multi-level, and fully reproducible evaluation framework. Under a unified dataset and scoring protocol, nine mainstream LLMs are systematically evaluated using standardized prompts, automated evaluation pipelines, semantic similarity matching, and weighted composite scoring. Results reveal critical limitations of current general-purpose models in construction automation tasks. By providing an open, standardized benchmark and automated evaluation workflow, this work lays a foundation for the reliable integration of LLMs into construction automation applications.

Fairness in machine learning remains a critical challenge, particularly in the presence of domain shift. We propose a unified fairness-aware framework for both domain generalization (DG) and unsupervised domain adaptation (UDA), which jointly addresses domain shift and sensitive-attribute bias through disentangled representation learning. The framework disentangles content, style, and sensitive factors, and uses them to generate augmented samples that reduce bias while maintaining predictive reliability. Extensive experiments on four datasets demonstrate that the proposed method achieves state-of-the-art performance in both DG and UDA settings. Moreover, it yields a stronger balance between classification accuracy and fairness across diverse domains and sensitive subgroups. By incorporating unlabeled target-domain data, our framework extends prior fairness-aware approaches that were limited to DG and provides new insight into fairness-aware learning under unsupervised adaptation. Overall, this work offers a practical step toward scalable and robust fairness-aware learning in multi-domain environments.

Large Language Models (LLMs) have demonstrated exceptional performance across a wide range of domains. However, inaccuracies in their outputs can lead to severe consequences in high-stakes areas such as finance and healthcare, where errors may result in the loss of money, time, or even lives. As a result, recent research has increasingly focused on uncertainty estimation in LLMs, aiming to quantify the trustworthiness of model-generated content given specific inputs. Despite this growing interest, the sources of uncertainty in LLMs remain insufficiently understood. As a result, this survey provides a comprehensive overview of uncertainty estimation for LLMs from the perspective of uncertainty sources, serving as a foundational resource for researchers entering the field. We begin by reviewing essential background on LLMs, followed by a detailed clarification of uncertainty sources relevant to them. We then introduce various uncertainty estimation methods, including both commonly used and LLM-specific approaches. Metrics for evaluating uncertainty are discussed, along with key application areas. Finally, we highlight major challenges and outline future research directions aimed at improving the trustworthiness and reliability of LLMs.

The Competitive Influence Maximization (CIM) problem involves multiple entities competing for influence in online social networks (OSNs). While Deep Reinforcement Learning (DRL) has shown promise, existing methods often assume users' opinions are binary and ignore their behavior and prior knowledge. We propose DRIM, a multi-dimensional uncertainty-aware DRL-based CIM framework that leverages Subjective Logic (SL) to model uncertainty in user opinions, preferences, and DRL decision-making. DRIM introduces an Uncertainty-based Opinion Model (UOM) for a more realistic representation of user uncertainty and optimizes seed selection for propagating true information while countering false information. In addition, it quantifies uncertainty in balancing exploration and exploitation. Results show that UOM significantly enhances true information spread and maintains influence against advanced false information strategies. DRIM-based CIM schemes outperform state-of-the-art methods by up to 57% and 88% in influence while being up to 48% and 77% faster. Sensitivity analysis indicates that higher network observability and greater information propagation boost performance, while high network activity mitigates the effect of users' initial biases.

Automatically solving optimization problems from natural language descriptions with both efficiency and reliability is highly desirable but remains challenging. Language model hallucinations and the limited availability of labeled datasets often result in misaligned formulations, code errors, and feasibility failures We propose UMCTS , an Uncertainty-aware Monte Carlo Tree Search framework that combines the language understanding capability of large language models with the reliability of well-established solvers. UMCTS structures the solution process into four stages: global instruction, assumptions, mathematical formulation, and solver code generation. It employs Monte Carlo Tree Search with semantic-equivalence pruning, prior-guided exploration, and solver-based feasibility checks. An LLM judge provides numerical reward signals, qualitative error information, and uncertainty estimates. These signals are backpropagated to guide the search and flag unreliable outputs. Across six public benchmarks, UMCTS achieves state-of-the-art solution accuracy, improves efficiency by reducing token usage.

Large Language Models (LLMs) have become increasingly pervasive, finding applications across many industries and disciplines. Ensuring the trustworthiness of LLM outputs is paramount, where Uncertainty Estimation (UE) plays a key role. In this work, a comprehensive empirical study is conducted to examine the robustness and effectiveness of diverse UE measures regarding aleatoric and epistemic uncertainty in LLMs. It involves twelve different UE methods and four generation quality metrics including LLMScore from LLM criticizers to evaluate the uncertainty of LLM-generated answers in Question-Answering (QA) tasks on both in-distribution (ID) and out-of-distribution (OOD) datasets. Our analysis reveals that information-based methods, which leverage token and sequence probabilities, perform exceptionally well in ID settings due to their alignment with the model's understanding of the data. Conversely, density-based methods and the P(True) metric exhibit superior performance in OOD contexts, highlighting their effectiveness in capturing the model's epistemic uncertainty. Semantic consistency methods, which assess variability in generated answers, show reliable performance across different datasets and generation metrics. These methods generally perform well but may not be optimal for every situation.

Uncertainty reasoning and quantification play a critical role in decision making across various domains, prompting increased attention from both academia and industry. As real-world applications become more complex and data-driven, effectively handling uncertainty becomes paramount for accurate and reliable decision making. This workshop focuses on the critical topics of uncertainty reasoning and quantification in decision making. It provides a platform for experts and researchers from diverse backgrounds to exchange ideas on cutting-edge techniques and challenges in this field. The interdisciplinary nature of uncertainty reasoning and quantification, spanning artificial intelligence, machine learning, statistics, risk analysis, and decision science, will be explored. The workshop aims to address the need for robust and interpretable methods for modeling and quantifying uncertainty, fostering reasoning decision-making in various domains. Participants will have the opportunity to share research findings and practical experiences, promoting collaboration and advancing decision-making practices under uncertainty.

This paper proposes U-FARE, an uncertainty-aware fair federated learning (FL) framework aimed at improving disease prediction in healthcare, with a specific focus on Alzheimer's disease detection. U-FARE incorporates evidential neural networks (ENN) to quantify uncertainty, enhancing both model fairness and accuracy. The framework ensures group-level fairness, providing consistent model performance across diverse healthcare environments despite data heterogeneity. We evaluate U-FARE on three real-world healthcare datasets-NACC, OASIS, and ADNI-comparing its performance to several state-of-the-art fairness-aware FL methods. Experimental results demonstrate that U-FARE outperforms baseline methods in both prediction accuracy and fairness, effectively balancing these two crucial aspects. The results also reveal the trade-off between fairness and accuracy, where higher fairness levels may compromise prediction accuracy. U-FARE achieves the highest accuracy (0.928) on the NACC dataset, consistently outperforms the competitive baseline q-FedAvg by 46%, particularly when higher fairness constraints are applied, and outperforms methods like Ditto and q-FFL with minimal accuracy variance and loss disparity. This is the first approach to simultaneously optimize fairness and accuracy in FL for Alzheimer's disease detection, providing a novel solution to the challenge of fair and effective AI in healthcare. The framework demonstrates the potential to address data heterogeneity while ensuring privacy and fairness in real-world applications.

Large Language Models (LLMs) are reshaping the landscape of computer science research, driving significant shifts in research priorities across diverse conferences and fields. This study provides a comprehensive analysis of the publication trend of LLM-related papers in 77 top-tier computer science conferences over the past six years (2019-2024). We approach this analysis from four distinct perspectives: (1) We investigate how LLM research is driving topic shifts within major conferences. (2) We adopt a topic modeling approach to identify various areas of LLM-related topic growth and reveal the topics of concern at different conferences. (3) We explore distinct contribution patterns of academic and industrial institutions. (4) We study the influence of national origins on LLM development trajectories. Synthesizing the findings from these diverse analytical angles, we derive ten key insights that illuminate the dynamics and evolution of the LLM research ecosystem.

Accurate quantification of both aleatoric and epistemic uncertainties is essential when deploying Graph Neural Networks (GNNs) in high-stakes applications such as drug discovery and financial fraud detection, where reliable predictions are critical. Although Evidential Deep Learning (EDL) efficiently quantifies uncertainty using a Dirichlet distribution over predictive probabilities, existing EDL-based GNN (EGNN) models require modifications to the network architecture and retraining, failing to take advantage of pre-trained models. We propose a plug-and-play framework for uncertainty quantification in GNNs that works with pre-trained models without the need for retraining. Our Evidential Probing Network (EPN) uses a lightweight Multi-Layer-Perceptron (MLP) head to extract evidence from learned representations, allowing efficient integration with various GNN architectures. We further introduce evidence-based regularization techniques, referred to as EPN-reg, to enhance the estimation of epistemic uncertainty with theoretical justifications. Extensive experiments demonstrate that the proposed EPN-reg achieves state-of-the-art performance in accurate and efficient uncertainty quantification, making it suitable for real-world deployment.

Large Language Models (LLMs) offer promising capabilities for tackling complex reasoning tasks, including optimization problems. However, existing methods either rely on prompt engineering, which leads to poor generalization across problem types, or require costly supervised training. We introduce SolverLLM, a training-free framework that leverages test-time scaling to solve diverse optimization problems. Rather than solving directly, SolverLLM generates mathematical formulations and translates them into solver-ready code, guided by a novel Monte Carlo Tree Search (MCTS) strategy. To enhance the search process, we modify classical MCTS with (1) dynamic expansion for adaptive formulation generation, (2) prompt backpropagation to guide exploration via outcome-driven feedback, and (3) uncertainty backpropagation to incorporate reward reliability into decision-making. Experiments on six standard benchmark datasets demonstrate that SolverLLM outperforms both prompt-based and learning-based baselines, achieving strong generalization without additional training.

The fusion of raw sensor data to create a Bird's Eye View (BEV) representation is critical for autonomous vehicle planning and control. Despite the growing interest in using deep learning models for BEV semantic segmentation, anticipating segmentation errors and enhancing the explainability of these models remain underexplored. This paper introduces a comprehensive benchmark for predictive uncertainty quantification in BEV segmentation, evaluating multiple uncertainty quantification methods across three popular datasets with three representative network architectures. Our study focuses on the effectiveness of quantified uncertainty in detecting misclassified and out-of-distribution (OOD) pixels while also improving model calibration. Through empirical analysis, we uncover challenges in existing uncertainty quantification methods and demonstrate the potential of evidential deep learning techniques, which capture both aleatoric and epistemic uncertainty. To address these challenges, we propose a novel loss function, Uncertainty-Focal-Cross-Entropy (UFCE), specifically designed for highly imbalanced data, along with a simple uncertainty-scaling regularization term that improves both uncertainty quantification and model calibration for BEV segmentation.

In real-world applications, data-driven classifiers often grapple with a three-pronged challenge: data arrives in a continuous stream, most data in the wild are often unlabeled, and there is a critical need to maintain fairness in predictions across different sub-groups. Existing methods falter when addressing all these three factors concurrently. This work tackles this challenge by addressing a novel paradigm: Fairness-Aware Active Online Learning. We introduce a simple yet effective approach - FACTION, which actively selects the most crucial data points for labeling, going beyond traditional methods by considering both model uncertainty (epistemic uncertainty) and a newly introduced fairness notion derived from this very uncertainty. Additionally, FACTION leverages a system adept at identifying out-of-distribution samples within online learning environ-ments. Extensive evaluations on real-world datasets, coupled with theoretical analysis, demonstrate FACTION's effectiveness in handling this complex challenge. Our model demonstrably outperforms relevant baselines adapted for this new setting.

Federated Learning (FL) offers a promising approach for collaborative model training in healthcare while preserving data privacy. However, existing FL methods often fall short in addressing two critical challenges: client-level fairness and compounded uncertainty from data heterogeneity and privacy-preserving mechanisms. We propose fair-LDP, a fairness-aware Local Differential Privacy framework that promotes fairness and privacy via uncertainty-guided aggregation in federated healthcare AI. fair-LDP leverages evidential neural networks (ENNs) to quantify predictive uncertainty and introduces a novel strategy that uses uncertainty-driven local differential privacy to guide fairness-aware updates while preserving data privacy. This ensures equitable performance across clients with varying data quality while mitigating the influence of unreliable or outlier updates. fair-LDP incorporates an adaptive mechanism that adjusts each client's privacy budget based on model performance, balancing fairness, privacy, and accuracy. We evaluate fair-LDP on real-world healthcare datasets under both IID and non-IID settings. Our experimental results show that it consistently outperforms state-of-the-art fairness-aware and privacy-preserving FL baselines, with no added computational overhead, while maintaining privacy guarantees comparable to homomorphic encryption and secure multiparty computation. By integrating uncertainty modeling, fairness-aware aggregation, and adaptive local differential privacy, fair-LDP provides a practical and principled solution for responsible, equitable, and privacy-preserving federated learning in healthcare.