During migration across instruction set architectures (ISAs), software package build repair is a critical task for ensuring the reliability of software deployment and the stability of modern operating systems. While Large Language Models (LLMs) have shown promise in tackling this challenge, prior work has primarily focused on single instruction set architecture (ISA) and homogeneous programming languages. To address this limitation, we introduce a new benchmark designed for software package build repair across diverse architectures and languages. Comprising 268 real-world software package build failures, the benchmark provides a standardized evaluation pipeline. We evaluate six state-of-the-art LLMs on the benchmark, and the results show that cross-ISA software package repair remains difficult and requires further advances. By systematically exposing this challenge, the benchmark establishes a foundation for advancing future methods aimed at improving software portability and bridging architectural gaps.

Theoretically, proper ensemble of multiple different models can improve the recommendation effectiveness. Unfortunately, most of the existing ensemble methods which depend on static aggregation strategies are not suitable for sequential recommendation, because they fail to distinguish the differences in predictive capabilities of base models among various users and contexts. In addition, most of the existing diversity measures, which are used in ensemble learning to strengthen the generalization performance, are designed for classification or regression tasks and not suitable for the ranking tasks in recommender systems. To tackle the above problems, we propose a Reinforcement Learning-based Adaptive Ensemble framework for sequential Recommendation (RLAERec), to effectively aggregate different base models for sequential recommendation. Specifically, we model the ensemble process as a Markov decision process (MDP) and design a dynamic weight allocation method according to constantly changing user behavior patterns and feedback. In addition, we propose an adaptive diversity measure for ranking ensemble learning, which can further enhance the ensemble performance. Extensive experiments on four real-world datasets demonstrate the effectiveness of the proposed model.

With the increasing adoption of hybrid cloud, existing authorization systems struggle to meet the increasingly complex authentication and authorization requirements in cloud environments. This paper proposes Casbin, a high-performance and user-friendly authorization framework. Casbin adopts a graph model-based policy language called PERM Modeling Language (PML), which supports various common authoriza tion models such as RBAC and ABAC. Casbin introduces multiple graph computation optimization techniques, enabling millisecond-level authorization decisions under a policy scale of millions. Furthermore, Casbin leverages Satisfiability Modulo Theories (SMT) solvers for automated formal verification of policies, improving policy reliability. Casbin also innovatively utilizes Graph Retrieval-Augmented Generation (Graph RAG) technology to assist in the semi-automated generation of high quality authorization policies. Experiments show that compared with methods like OpenFGA, Cedar, and SpiceDB, Casbin's average policy evaluation latency is reduced by 81.7%, and throughput is increased by 4.6 times. On large-scale complex policy sets, Casbin's SMT-based formal verification achieves a 92.1% recall rate and 94.5% precision rate for violation detection, with verification time controlled within 143.2 seconds. User studies indicate that compared to OpenFGA, Cedar, XACML, etc., PML's policy completion time is shortened by 57.3%, and accuracy is increased by 16.4%.

Online service systems operate under high concurrency, dynamic workloads, and distributed execution environments, which increase their susceptibility to faults and performance degradation. To ensure stable operation, these systems commonly monitor performance metrics, where timely anomaly detection enables early intervention and fault prevention. Existing metrics-based anomaly detection methods are typically unsupervised, reconstructing unlabeled data to model normal patterns and detecting anomalies as deviations from these patterns. However, they often overlook the presence of unlabeled anomalous data (noise) within the training set, which hinders the model's ability to accurately learn normal behavior and degrades detection accuracy. To address this limitation, we propose a noise-robust, representation-enhanced reconstruction-based anomaly detection method for metrics. The method trains a reconstruction model to capture normal time-series patterns and identifies noisy samples based on their deviation from the latent representation center, assigning them lower training weights. Both the reconstruction outputs and the representation centers jointly define the normal pattern of each sample, reducing the influence of anomalous noise during modeling. Extensive experiments show that our method achieves an F1 score of 97.00% on four widely used datasets, surpassing existing state-of-the-art approaches. Ablation studies further confirm the effectiveness of each module.

Objective:Unplanned reoperations are widely recognised as key indicators of surgical quality and patient safety. Analysing the underlying causes of such reoperations is critical for guiding clinical decision-making and optimising patient management. Most existing studies relied on electronic medical records (EMRs) for simple retrospective analyses and focused exclusively on single-disease cohorts, limiting their generalisability. Tools capable of predicting the causes of unplanned reoperations using multimodal EMRs across diverse clinical settings remain scarce. This study aimed to develop an artificial intelligence (AI)-based multimodal system integrating structured and unstructured EMR data across institutions. Methods and analysis:We developed the Multi-modal Prediction System for Causes of Unplanned Reoperation (MPSUR), an AI-based framework trained on a retrospective cohort of 2922 cases collected from 15 departments across 8 hospitals (2015-2024). The system integrates structured series data (such as age and sex) and clinical text (such as diagnoses and procedures) using the Graph Convolutional Network and Time and Frequency Recurrent Neural Network. Modality-specific and fused predictions were generated using linear classifiers. Results:In the clinical reader study, the MPSUR outperformed surgeons across all departments (mean accuracy=60.27%, 95% CI 58.27% to 62.26%). Feature ablation showed that variables such as procedure and department were the most predictive, with their removal significantly reducing performance (-3.05%, p<0.01) for the internal dataset and (-3.42%, p<0.005) for the external dataset. MPSUR achieved 62.03% and 56.41% accuracy on internal and external datasets, respectively, outperforming classical baselines by up to 8%. The SD across cross-validation folds was minimal, underscoring model stability. It also remained robust when only text or series data were available. Conclusion:The MPSUR provides an automatic, accurate, interpretable and generalisable tool for predicting the causes of unplanned reoperations using multimodal EMRs. Its strong performance across internal and external datasets, as well as its superiority over clinicians, supports its clinical utility as a decision-support system to enhance surgical safety and patient outcomes.

Although Multimodal Large Language Models have achieved strong performance across a wide range of vision-language tasks, they still suffer from hallucinations, where model outputs become inconsistent with the visual content, textual context, or commonsense knowledge. Existing studies primarily address this problem through coarse-grained detection. However, these approaches often provide insufficient diagnostic information for understanding hallucination types and supporting downstream hallucination mitigation. To bridge this gap, we propose fine-grained hallucination diagnosis for MLLMs, a new unified task that jointly performs hallucination detection, classification, and interpretable explanation generation. We develop an automated data generation pipeline and construct HalluScope-30K, a large-scale diagnostic dataset covering eight sources and five task categories. Based on this dataset, we design a multi-granular joint reward function and train two diagnosis models, HalluScope-4B and HalluScope-8B, which achieve state-of-the-art performance on both the MHALO benchmark and our fine-grained hallucination classification benchmark. Notably, detection and classification are mutually beneficial under joint optimization. Furthermore, diagnosis-driven feedback experiments show that the fine-grained diagnostic explanations produced by our model effectively guide target models to correct their hallucinations, with full diagnosis substantially outperforming all baselines on both Qwen3-VL-8B-Instruct and LLaVA-1.5-7B.

Information-seeking agents have emerged as a powerful paradigm for solving knowledge-intensive tasks. Existing information-seeking agents are typically specialized for open web, documents, or local knowledge bases, which constrains scalability and cross-domain generalization. In this work, we investigate how to consolidate heterogeneous information-seeking agents into a single foundation agentic model. We study two complementary consolidation strategies: data-level consolidation, which jointly trains a unified model on a mixture of domain-specific datasets, and parameter-level consolidation, which merges independently trained agent models at the parameter level. Our analysis compares these approaches in terms of performance retention, cross-domain generalization, and interference across information-seeking behaviors. Our results show that data-level consolidation remains a strong and stable baseline, while parameter-level consolidation offers a promising, efficient alternative but suffers from interference and robustness challenges. We further identify key design factors for effective agent consolidation at the parameter level, including fine-grained merging granularity, awareness of task heterogeneity, and principled consensus strategy.

Drug combination recommendation aims to provide personalized medication prescriptions for patients according to their historical visit records and current condition. Several recent studies try to utilize graph structures to better model the relationships among diagnoses, procedures, and drugs for drug combination recommendation. However, they primarily focus on pairwise relationships and fail to capture the higher-order connections and combination effects among different medical entities. Additionally, existing approaches often overlook the multifaceted nature of diagnoses and procedures within patients’ medical records. To address these issues, we propose a hierarchical hypergraph convolution network for drug combination recommendation. Specifically, we design an innovative hierarchical hypergraph structure and corresponding convolution network to capture the complex, higher-order relationships among medical entities. Furthermore, we design a capsule-based representation learning method to model the multifaceted nature of patients’ medical records for drug combination recommendation. Extensive experiments on several real-world datasets of different hospital departments demonstrate the effectiveness of the proposed approach compared with state-of-the-art methods.

Medication recommendation systems aim to provide personalized and safe medication options based on individual patient records. However, existing approaches often face challenges related to inadequate modeling of complex relationships within Electronic Health Records (EHRs), data sparsity, and a lack of explainability for recommendations. In this paper, we present a Knowledge-enhanced Explainable HyperGraph Convolution Network (KEHGCN) that constructs a hierarchical hypergraph structure to capture the multi-level relationships within EHR data. By incorporating external knowledge graphs, our approach introduces additional positive relations that help alleviate the impact of data sparsity on model learning. Furthermore, by performing generalized metapath construction and selection on the knowledge graph, our approach achieves effective knowledge filtering and extracts semantically meaningful metapaths, thereby further enhancing the explainability of the recommendation results. We also explicitly introduce negative relations present in the domain knowledge to improve the safety of medication recommendation. Extensive experiments on different hospital departments of MIMIC-III and MIMIC-IV datasets demonstrate that KEHGCN outperforms other state-of-the-art baselines.

Root cause Localization methods play a crucial role in ensuring the stability of large-scale microservice systems. However, existing methods either rely on unsupervised approaches with limited localization accuracy or on supervised learning that requires large volumes of historical anomaly data. Such data is often unavailable in newly deployed systems. To address this limitation, we attempt to use the labeled data from mature systems to help new systems build root cause localization models. Specifically, we propose UDA-RCL, an unsupervised domain adaptation root cause localization method using multimodal data (log, metric and trace data). UDA-RCL first incorporates an aggregation based event extraction module to standardize the format of multimodal data from different systems. Then, it utilizes a multimodal event encoder and multimodal domain adversarial adaptation module to narrow the feature distribution gap between different systems. Furthermore, taking into account the situation of sparse anomaly samples, existing methods' classifiers are either hard to transfer to other systems or struggle to capture the process of anomaly propagation, we propose a PageRank classifier module. This module employs a neural network embedded with anomaly propagation rules to output the final root cause ranking results, alleviating the issue of sparse anomaly samples. Extensive experiments have proven that our method achieves the best results in both supervised and transfer learning scenarios.

In today’s economic climate, enterprises face a variety of internal and external risks, making bankruptcy prediction critical for risk management. The traditional statistical and machine learning-based methods mainly rely on economic indicators, which are insufficient for deducing the risk propagation among enterprises. Recently, researchers have begun to explore the use of graph neural networks, utilizing their message-passing mechanisms to simulate the risk propagation process. However, existing graph-based methods often neglect degree imbalances, leading to high misjudgment rates for sparsely connected nodes. Furthermore, existing methods typically use a risk-oriented decision model to evaluate the likelihood of bankruptcy, which may lead to the overestimation of bankruptcy probabilities.To address these issues, we propose a novel bankruptcy prediction model which consists of several key components, including a data-driven explicit risk encoding module, a global multihead attention-based implicit risk encoding module, a hierarchical hypergraph-based external risk enhancement module, and a dual-decision expert-based risk assessment module. We extend the traditional graph structure to a hierarchical hypergraph structure and design a corresponding information propagation strategy to alleviate the degree imbalance issue. Furthermore, a dual-decision assessment module is designed to integrate the perspectives of both risk and non-risk experts to prevent the overestimation of bankruptcy probabilities. Extensive experiments conducted on a real-world dataset demonstrate the effectiveness of the proposed model, which achieves an accuracy of 77.64% and an AUC of 0.8270, significantly outperforming existing methods.

Previous LLMs-based RL studies typically follow either supervised learning with high annotation costs, or unsupervised paradigms using voting or entropy-based rewards. However, their performance remains far from satisfactory due to the substantial annotation cost and issues such as model collapse or reward hacking. To address these issues, we introduce a new perspective inspired by cognitive learning theory and propose a novel approach called EasyRL. The core of EasyRL is to simulate the human cognitive acquisition curve by integrating reliable knowledge transfer from easy labeled data with a progressive divide-and-conquer strategy that tackles increasingly difficult unlabeled data. Specifically, we initialize a warm-up model using supervised RL with few-shot labeled data. This is followed by a divide-and-conquer pseudo-labeling strategy on difficult unlabeled data, combining consistency-based selection for low-uncertainty cases and reflection-based resolution for medium-uncertainty cases. Finally, difficulty-progressive self-training with iterative pseudo-labeling and RL further strengthens the model's reasoning capability. EasyRL provides a unified self-evolving framework that facilitates data-efficient post-training of LLMs. Experimental results on mathematical and scientific benchmarks demonstrate that EasyRL, using only 10

Diffusion models enable high-fidelity image editing but can also be misused for unauthorized style imitation and harmful content generation. To mitigate these risks, proactive image protection methods embed small, often imperceptible adversarial perturbations into images before sharing to disrupt downstream editing or fine-tuning. However, in realistic post-release scenarios, content owners cannot control downstream processing pipelines, and protections optimized for a surrogate model may fail when attackers use mismatched diffusion pipelines. Existing purification methods can weaken protections but often sacrifice image quality and rarely examine architectural mismatch. We introduce a unified post-release purification framework to evaluate protection survivability under model mismatch. We propose two practical purifiers: VAE-Trans, which corrects protected images via latent-space projection, and EditorClean, which performs instruction-guided reconstruction with a Diffusion Transformer to exploit architectural heterogeneity. Both operate without access to protected images or defense internals. Across 2,100 editing tasks and six representative protection methods, EditorClean consistently restores editability. Compared to protected inputs, it improves PSNR by 3-6 dB and reduces FID by 50-70 percent on downstream edits, while outperforming prior purification baselines by about 2 dB PSNR and 30 percent lower FID. Our results reveal a purify-once, edit-freely failure mode: once purification succeeds, the protective signal is largely removed, enabling unrestricted editing. This highlights the need to evaluate protections under model mismatch and design defenses robust to heterogeneous attackers.

The development of artificial intelligence and medical informatics has empowered the medication recommendation systems with enhanced capabilities. However, existing methods struggle with the data imbalance problem in Electronic Health Records (EHRs), where the majority of records are concentrated on a limited subset of common diagnoses, procedures, and medications. It hampers the models’ ability to recommend appropriate medications when dealing with uncommon or multifaceted cases. In addition, existing approaches often fail to adequately model the complex relationships inherent in heterogeneous medical data sources, especially medication molecular structure information. This gap restricts the potential for uncovering meaningful associations among diverse clinical entities. To address these issues, we design a hierarchical attention-based pretraining strategy, leveraging the semantic hierarchies of medical entity codes to facilitate knowledge transfer, so as to alleviate the challenge of data imbalance. Furthermore, we design a novel bilateral hierarchical heterogeneous hypergraph convolution network for medication recommendation. Specifically, we construct specialized hypergraphs for both EHR data and medication molecular structure data, enabling hypergraph convolution to capture high-order relationships while promoting bilateral knowledge enhancement between these heterogeneous data sources. This comprehensive integration allows the model to effectively capture the relationships among clinical and molecular information. Experimental results on different hospital departments of MIMIC-III and MIMIC-IV datasets demonstrate the superior performance of our model compared to state-of-the-art methods. Our source code is released at: https://github.com/LusiaZ/BH3-MedRec .

Enterprise bankruptcy prediction is essential for financial investment and corporate management. Traditional machine learning based methods mainly focus on internal risk mining while ignoring the external risks propagated among enterprises. Recent research efforts have attempted to use graph neural networks to model external risks. However, these methods generally lack interpretability and suffer from transferability. To address these issues, we propose a novel enterprise bankruptcy prediction model with meta-path denoising and capsule network modeling. Specifically, we design an automatic meta-path generation and selection method to improve the model’s transferability while minimizing the information loss and noise introduction. Furthermore, a hierarchical meta-path information aggregation method is designed to enhance the model’s interpretability, and a capsule network based risk assessment module is designed to dynamically capture enterprises’ risky and non-risky factors for better bankruptcy prediction. Extensive experiments demonstrate the effectiveness and interpretability of the proposed model.

Reinforcement learning (RL) has emerged as a powerful post-training paradigm for enhancing the reasoning capabilities of large language models (LLMs). However, reinforcement learning for LLMs faces substantial data scarcity challenges, including the limited availability of high-quality external supervision and the constrained volume of model-generated experience. These limitations make data-efficient reinforcement learning a critical research direction. In this survey, we present the first systematic review of reinforcement learning for LLMs under data scarcity. We propose a bottom-up hierarchical framework built around three complementary perspectives: the data-centric perspective, the training-centric perspective, and the framework-centric perspective. We develop a taxonomy of existing methods, summarize representative approaches in each category, and analyze their strengths and limitations. Our taxonomy aims to provide a clear conceptual foundation for understanding the design space of data-efficient RL for LLMs and to guide researchers working in this emerging area. We hope this survey offers a comprehensive roadmap for future research and inspires new directions toward more efficient and scalable reinforcement learning post-training for LLMs.

LLMs are widely deployed through cloud-hosted inference services, where Just-in-Time (JIT) compilation is used to reduce recurring framework and GPU-launch overhead. JIT serving introduces a host-side control plane that selects compiled artifacts and orchestrates their execution on the GPU. Meanwhile, the shared cloud setting has motivated a growing body of bit-flip attacks (BFAs) against LLM/DNN inference. Most existing BFAs target model parameters or weights and require model-specific knowledge. A smaller body of work reduces this dependency by faulting executable code, yet still corrupts code that directly implements model computation, limiting their attack effect to inference depletion. We present JITterFlip, the first BFA targeting the host-side JIT serving control plane of GPU-based LLM inference. By faulting CPU-resident serving decisions rather than model computation, JITterFlip enables both gibberish output generation and a correct-output sponge attack. To identify exploitable targets in a large JIT compiler stack, JITterFlip develops a decision-guided fault-vulnerable code analysis. Across four text and multimodal LLM workloads, the identified vulnerable code faults exhibit cross-model transferability, produce gibberish outputs with PPL ratios of 15.45× to 2.48×10^6×, and demonstrate correct-output sponge attacks with latency amplification of 2.03× to 181.90×. JITterFlip also bypasses recent BFA defenses for LLMs while retaining both attack effects. Last, we demonstrate end-to-end Rowhammer attacks across four LLMs: a single bit flip in CPU-resident branch code propagates across the CPU-GPU boundary to disrupt GPU-executed inference without direct access to GPU memory, reaching up to 7.23×10^6× PPL amplification or 124.97× latency amplification while preserving the exact generated output.

As modern microservice systems grow increasingly popular and complex-often consisting of hundreds or even thousands of fine-grained, interdependent components-they are becoming more susceptible to frequent and subtle failures. Ensuring system reliability therefore hinges on accurate and efficient failure localization. Traditional failure localization approaches based on small models lack the flexibility to adapt to diverse failure scenarios, while recent LLM-based methods suffer from two major limitations: they often rely on rigid invocation workflows that constrain the model's ability to dynamically explore optimal localization paths, and they require resource-intensive inference, making them cost-prohibitive for real-world deployment. To address these challenges, we explore the use of reinforcement fine-tuning to equip lightweight LLMs with reasoning and self-refinement capabilities, significantly improving the cost-effectiveness and adaptability of LLM-based failure localization. We begin with an empirical study to identify three key capabilities essential for accurate localization. Building on these insights, we propose a progressive multi-stage GRPO fine-tuning framework, which integrates a multi-factor failure localization grader and a recursion-of-thought actor module. The resulting model, ThinkFL, not only outperforms existing state-of-the-art LLMs and baseline methods in localization accuracy but also reduces end-to-end localization latency from minutes to seconds, demonstrating strong potential for real-world applications.

The current paradigm of training large language models (LLMs) on publicly available Web data is becoming unsustainable, with high-quality data sources in specialized domains nearing exhaustion. Federated Learning (FL) emerges as a practical solution for the next generation of AI on a decentralized Web, enabling privacy-preserving collaborative fine-tuning by leveraging private data distributed across a global client base. While Low-Rank Adaptation (LoRA) is the standard for efficient fine-tuning, its application in federated settings presents a critical challenge: communication overhead remains a significant bottleneck across the Web's heterogeneous network conditions. The structural redundancy within LoRA parameters not only incurs a heavy communication burden but also introduces conflicts when aggregating client updates. To address this, we propose FedSRD, a Sparsify-Reconstruct-Decompose framework designed for communication-efficient FL. We first introduce an importance-aware sparsification method that preserves the structural integrity of LoRA updates to reduce the uploaded parameter count. The server then reconstructs and aggregates these updates in a full-rank space to mitigate conflicts. Finally, it decomposes the global update into a sparse low-rank format for broadcast, ensuring a symmetrically efficient cycle. We also propose an efficient variant, FedSRD-e, to reduce computational overhead. Experimental results on 10 benchmarks demonstrate that our framework significantly reduces communication costs by up to 90\% while even improving model performance on heterogeneous client data.

Identity-based encryption with equality test (IBEET) extends identity-based encryption to support the comparison between messages in two distinct ciphertexts associated with different identities. Nevertheless, nearly all of the previous IBEET schemes have the problem that once the trapdoor is sent to the cloud server, it becomes uncontrollable. Meanwhile, the only existing IBEET scheme coping with this problem, namely IBEET with datestamp-based authorization (IBEET-DBA), suffers from the following flaws. (1) A trapdoor merely expresses a single time interval, which is inexpressive. (2) It is inflexible due to only supporting limited types of authorization. (3) It is merely proven secure in the random oracle model, which has been pointed out that it may lead to security risks in practicality. Considering the aforementioned problems, we propose an IBEET with flexible time-based authorization (IBEET-FTBA) scheme, supporting wildcards and putting no constraint on authorization types. We prove our scheme secure against adaptive chosen-ciphertext attacks in the standard model. Subsequently, we give a detailed theoretical analysis of our IBEET-FTBA construction and the main existing IBEET ones. Furthermore, we implement ours and the IBEET-DBA construction as a benchmark, and conduct numerous experiments. Both results demonstrate that ours has better performance in the scenario of multiple time intervals.