Search Results (252)

Large Language Models (LLMs) have become foundational to modern Artificial Intelligence (AI), enabling advanced reasoning, multimodal understanding, and scalable human-AI interaction across diverse domains. This survey provides a comprehensive review of major proprietary and open-source LLM families, including GPT, LLaMA 2, Gemini, Claude, DeepSeek, Falcon, and Qwen. It systematically examines architectural advancements such as transformer refinements, mixture-of-experts paradigms, attention optimization, long-context modeling, and multimodal integration. The paper further analyzes alignment and safety mechanisms, encompassing instruction tuning, reinforcement learning from human feedback, and constitutional frameworks, and discusses their implications for controllability, reliability, and responsible deployment. Comparative analysis of training strategies, data curation practices, efficiency optimizations, and application settings highlights key trade-offs among scalability, performance, interpretability, and ethical considerations. Beyond synthesis, the survey introduces a structured taxonomy and a feature-driven comparative study of over 50 reconstructed LLM architectures, complemented by an interactive visualization interface and an open-source implementation to support transparency and reproducibility. Finally, it outlines open challenges and future research directions related to transparency, computational cost, data governance, and societal impact, offering a unified reference for researchers and practitioners developing large-scale AI systems. Full article

Docker systems are gaining widespread use due to their consistency, scalability, and ease of application portability, which addresses specific security challenges. Traditional monitoring and intrusion detection systems based on predefined rules often struggle with advanced attack patterns due to a lack of the capability to correlate incoming log messages. This paper proposes a correlation-aware log analysis approach based on a Mixture-of-Experts (MoE) large language models, enabling detection of malicious activity by analyzing both individual log entries and their contextual relationships within sequences of logs. The system processes each log in the context of 50 preceding messages, allowing identification of attack patterns that are not observable from isolated logs. To evaluate the approach, we generated a comprehensive dataset based on OWASP Top 10 attack scenarios, enriched with zero-day attacks such as Log4j and React2Shell, deployed in a distributed Docker Swarm environment. Multiple LLMs were evaluated under identical hardware conditions to ensure fair comparison. Experimental results demonstrate that while most models achieve comparable performance on single-log detection, significant differences emerge in contextual analysis. The proposed MoE-based approach demonstrates superior effectiveness, achieving an F1 score from 0.993 to 0.998 for contextual-log analysis. The contribution of this research is the novel use of MoE LLMs for log analysis, the distinct novel attack log dataset, and the unique framework based on offline technology that conserves hardware resources and data privacy. Full article

The integration of artificial intelligence in healthcare has revolutionized research, diagnostics, and patient care. In particular, the emergence of ChatGPT and the recent rise of DeepSeek have drawn significant attention to this landscape. This empirical work provides a comparative performance evaluation of these models and their role in medical education from a privacy perspective. Our study shows that, at present, DeepSeek-R1 is based on enhanced reasoning capabilities, while the ChatGPT-4o counterpart, on the other hand, is designed with support for text, audio, and vision which is lacking in standard DeepSeek models. However, due to the distillation techniques used, DeepSeek has an advantage over ChatGPT in terms of resource requirements. We evaluate both models on the MedQA dataset from privacy perspectives where ChatGPT returns 94% correct answers compared to 91% with DeepSeek, though DeepSeek demonstrates consistency in explanatory responses. Full article

Several highway agencies have implemented full-depth reclamation (FDR) as a sustainable technology for rehabilitating deteriorated asphalt pavements. However, the lack of standardized mix design procedures and limited field assessment, in terms of rutting and cracking resistance, pose challenges to the widespread implementation of FDR. This study addresses these challenges by synthesizing current FDR mix design and construction practices and validating highway agency-recommended practices through laboratory performance evaluation. The study objectives were achieved by (1) reviewing current FDR mix design and construction specifications of highway agencies across the US and internationally, (2) conducting surveys with highway agencies and interviews with subject matter experts (SMEs), and (3) evaluating the laboratory performance of FDR mixtures. Based on the findings from the literature, survey responses, and SME interviews, three FDR mixtures were designed in the lab: (i) cement-only, (ii) asphalt emulsion and cement, and (iii) foamed asphalt and cement. Each mix was then evaluated for rutting susceptibility using the Asphalt Pavement Analyzer (APA) and cracking resistance using the indirect tensile (IDT) test to identify optimum dosages of bituminous and cementitious additives. Laboratory results showed that FDR mixtures with 3% asphalt emulsion and 1% cement improved rutting resistance by 46% and cracking performance by 70% compared to cement-only mixtures with 4% cement. In contrast, foamed asphalt did not result in a significant improvement in FDR performance. Survey responses indicated that 89% of respondents reported good field performance of FDR, with Pennsylvania and North Dakota exhibiting excellent performance 10 years after construction. Full article

Few-Shot Class-Incremental Learning (FSCIL) aims to sequentially learn new classes from very few labelled samples while preventing the forgetting of previously acquired knowledge, which has important practical value for remote sensing scene classification (RSSC). Recent studies have shown that applying a Vision Transformer (ViT) pre-trained on natural image datasets to FSCIL tasks can achieve significantly superior performance. Nevertheless, a substantial domain distribution gap exists between natural images and remote sensing images, which leads to severe performance degradation when such models are directly transferred to RSSC. To address the domain gap alongside FSCIL’s inherent stability–plasticity dilemma and overfitting under data scarcity, we propose a Dynamic Expansion Mixture-of-Experts with Pre-trained Vision Transformer (DEM-ViT) framework. Specifically, to alleviate the domain discrepancy, DEM-ViT incorporates an Adapter-Based Mixture-of-Experts (ABMoE) module, which captures the diverse visual patterns of remote sensing scenes through feature reconstruction in the representation space and collaborative learning among multiple experts. Furthermore, to address the stability–plasticity dilemma in FSCIL, we propose a Dynamic Expert Expansion (DEE) strategy, which progressively expands the model capacity along the incremental sessions. DEE provides sufficient space for learning new knowledge while mitigating the forgetting of previous knowledge. In addition, we propose a Semantic-Guided Feature Alignment (SGFA) method to reduce the risk of overfitting under data-scarce conditions. SGFA leverages textual information to construct robust text prototypes and uses them to calibrate the visual feature space. Extensive experiments across three benchmarks indicate that our framework exhibits highly competitive performance compared with state-of-the-art methods. Full article

Scientific publishing is currently constrained by an unstructured narrative bottleneck paradigm, which increasingly diverges from the scale, complexity, and computational nature of modern research. Despite rapid advancements in data generation and analysis, scientific knowledge is predominantly disseminated as static narrative artifacts, thereby limiting reproducibility, machine accessibility, and cumulative integration. This study explores how scientific communication can be restructured to facilitate scalable validation and reliable knowledge accumulation. We propose the Object-Oriented Scientific Information paradigm, wherein scientific contributions are represented as executable, machine-interpretable objects that integrate structured data, reproducible methodologies, and formally encoded semantic claims. To operationalize this paradigm, we delineate the architecture of an Autonomous Knowledge Engine, a modular neuro-symbolic system that combines domain-specialized Mixture-of-Experts routing, formal verification of claims, and an information-theoretic filter based on marginal information gain. This architecture enables continuous validation, redundancy control, and the integration of scientific contributions within an active knowledge graph. The analysis demonstrates that Object-Oriented Scientific Information (OOSI) and Autonomous Knowledge Engine (AKE) fundamentally differ from existing document-based, executable, and semantic publishing models by shifting epistemic control from narrative evaluation to computational verification. We conclude that transitioning toward a computable scientific record is essential for sustaining reliable and self-correcting science in the context of accelerating knowledge production. Full article

Industrial pouring processes operate under highly dynamic conditions where small deviations can lead to defects, scrap, and production losses. Although modern foundries are equipped with multiple sensors and visual inspection systems, most monitoring approaches remain fragmented, unimodal, and difficult to interpret. Furthermore, annotated anomalous samples in industrial settings are scarce, hindering the development of traditional methods. As a result, many critical pouring anomalies are detected too late or lack sufficient contextual information for effective decision making. In this work, we propose a multimodal framework for industrial scene characterisation that combines visual information and process signals through an explainable Mixture-of-Experts (MoE)-style expert-fusion strategy. First, we deploy an ensemble of specialised modules that collaborate to identify regions of interest, assess pouring quality, and contextualise events within the production process, thereby generating an interpretable description of pouring events. Second, we introduce a novel anomaly detection method for multimodal video data, combining a self-supervised transformer with an outlier-aware clustering algorithm. Our approach effectively identifies rare anomalies without requiring extensive manual labelling. The resulting information is structured into a digital twin-ready representation, supporting synchronisation between the physical system and its virtual counterpart. This solution provides a scalable, deployable pathway to transform heterogeneous industrial data into actionable knowledge, supporting advanced monitoring, anomaly detection, and quality control in real foundry environments. Full article

Multispectral remote sensing image super-resolution (RSISR) aims to reconstruct high-frequency details while preserving cross-band structural consistency under strict computational budgets. However, real-world satellite imagery exhibits heterogeneous distortions, ranging from band-dependent noise to spatially varying texture degradation, rendering uniform restoration strategies suboptimal. To address these challenges, we propose a unified framework that integrates cue extraction, expert specialization, and efficiency-aware restoration. Specifically, a Distortion-Aware Feature Extractor (DAFE) explicitly encodes distortion cues by synthesizing fixed frequency bases, learnable residual components, lightweight spatial edge representations, and noise proxies. Subsequently, a Distortion-Aware Expert Choice (DAEC) router utilizes these cues to establish distortion-conditioned affinities and performs capacity-constrained, load-balanced expert assignment. Finally, a parameter-shared Mixture-of-Experts (PS-MoE) architecture employs shared expert parameters across spectral bands, augmented by band-wise low-rank adapters, to enable coarse-to-fine restoration with minimal computational overhead. Extensive experiments on the SEN2VENμS and OLI2MSI datasets demonstrate that the proposed method achieves a PSNR of 49.38 dB on SEN2VENμS 2×, 45.91 dB on SEN2VENμS 4×, and 45.94 dB on OLI2MSI 3×. Compared to the strongest baseline for each task, our method yields PSNR improvements of 0.12 dB, 0.10 dB, and 0.09 dB, respectively, while simultaneously reducing FLOPs and parameter counts. These results confirm that explicit distortion modeling and parameter-shared expert specialization provide an effective and computationally efficient solution for multispectral remote sensing image super-resolution. Full article

Remote sensing images are highly susceptible to spatially non-uniform haze under complex atmospheric conditions, leading to contrast degradation and structural detail loss. Moreover, remote sensing scenes usually exhibit complex spatial structures, highly uneven haze distribution, and significant statistical variability, which further increases the difficulty of haze removal. To address this issue, we revisit the haze degradation mechanism of remote sensing imagery and propose a dynamic histogram-sorted Transformer dehazing method from the perspectives of statistical distribution modeling and region-adaptive restoration. Specifically, a Histogram-Sorted Adaptive Attention is designed to map spatial features into the statistical distribution domain through a dynamic histogram sorting mechanism, enabling explicit discrimination and precise modeling of regions with different haze densities. Meanwhile, a Perception-Adaptive Feed-Forward Network is constructed, which incorporates a stable routing-based mixture-of-experts mechanism to adaptively select restoration strategies according to local texture characteristics and global haze density, thereby significantly enhancing the adaptability of the model in complex remote sensing scenarios. Extensive experimental results demonstrate that the proposed method achieves superior performance over existing approaches across multiple remote sensing benchmark datasets, effectively improving both visual quality and robustness of remote sensing imagery. Full article

The precise identification of minute, high-velocity entities within unconstrained visual fields represents a significant hurdle in computational perception. This difficulty primarily arises from the geometric degradation stemming from scale volatility, motion-induced asymmetry, and heterogeneous background clutter. To mitigate the critical deficit of high-fidelity benchmarks for dynamic micro-targets, we present Soccer-Wild. This comprehensive dataset is characterized by the extreme visual complexity of microscopic objects in diverse ecological settings. Built upon this empirical foundation, we introduce GOAL (Global Object Alignment for Localization). This novel computational paradigm is designed to enhance the weak features of tiny targets by integrating frequency-domain filtering, dynamic feature routing, and entropy-guided probabilistic modeling. The GOAL framework rigorously preserves spatial-structural equilibrium and information fidelity through three synergetic mechanisms: (1) Spectral Purification: We implement a Frequency-aware Spectral Gating approach that operates in the Fourier manifold, suppressing stochastic noise to accentuate the spectral signatures of the targets; (2) Geometric Adaptation: A Multi-Granularity Mixture of Experts (MG-MoE) is formulated with heterogeneous receptive fields to dynamically rectify anisotropic distortions caused by kinetic blurring. This adaptive routing ensures cross-state representation consistency; (3) Information Recovery: We propose Information-Guided Gaussian Distribution Estimation (IGDE), which utilizes information entropy to conceptualize target coordinates as radially symmetric probability densities. This facilitates the implicit recovery of latent signals typically discarded by rigid deterministic regression. Empirical validations on the Soccer-Wild and VisDrone2019 benchmarks reveal that the proposed methodology yields substantial gains in precision. Specifically, our model achieves 40.0% and 40.4% AP (Average Precision), respectively, establishing a new state-of-the-art for localizing highly dynamic, micro-scale objects. Full article

Graph Neural Networks (GNNs) have emerged as the standard for learning representations from graph-structured data. While traditional architectures relying on message-passing mechanisms excel in homophilic settings, they essentially function as fixed low-pass filters. However, this smoothing operation limits their ability to generalize to heterophilic graphs, where connected nodes often exhibit dissimilar labels and high-frequency signals are crucial for discrimination. Furthermore, existing Mixture-of-Experts (MoE) methods for graphs often suffer from local-view routing, failing to capture global structural context during expert selection. To address these challenges, this paper proposes SS-AdaMoE, a novel Spatio-Spectral Adaptive Mixture of Experts framework designed for robust node classification across diverse graph patterns. Specifically, a Dual-Domain Expert System is constructed, integrating heterogeneous spatial aggregators with learnable spectral filters based on Bernstein polynomials. This allows the model to adaptively capture arbitrary frequency responses—including high-pass and band-pass signals—which are overlooked by standard GNNs. To resolve the locality bias, a Hierarchical Global-Prior Gating Network augmented by a Linear Graph Transformer is introduced, ensuring that expert selection is guided by both local node features and global topological awareness. Extensive experiments are conducted on five benchmark datasets spanning both homophilic and heterophilic networks. The results demonstrate that SS-AdaMoE consistently outperforms baselines, achieving accuracy improvements of up to 2.65% on Chameleon and 1.41% on Roman-empire over the strongest MoE baseline, while surpassing traditional GCN architectures by margins exceeding 28% on heterophilic datasets such as Texas. These findings validate that the synergy of learnable spectral priors and global gating effectively bridges the gap between spatial aggregation and spectral filtering. Full article

Bundle recommendation has gained significant attention, but it faces two key challenges: sparse interaction data and complex UB, UI, and BI relations. Recent work uses multi-view contrastive learning, yet current frameworks rely on fixed-weight fusion that ignores view-specific importance and suffers from gradient suppression on sparse data. We propose TriadCBR, a tri-view adaptive contrastive learning architecture for bundle recommendation. It uses a simplified GCN to learn view-specific representations and a Mixture-of-Experts (MoE) module to generate personalized fusion weights, addressing the limitations of fixed-weight fusion. TriadCBR further incorporates a fine-grained contrastive module integrating InfoNCE, DCL, and Barlow Twins. This combination effectively mitigates gradient vanishing from invalid negatives and minimizes cross-view feature redundancy. To handle data sparsity, we design a Difficulty-Aware BPR (DA-BPR) with curriculum augmentation to dynamically refine the ranking trajectory. Extensive experiments on Youshu, iFashion, and NetEase demonstrate that TriadCBR achieves statistically significant improvements, boosting Recall and NDCG by an average of 3.61%, with 9 of 12 metric–dataset combinations reaching statistical significance, over state-of-the-art baselines, validating the robustness of its dynamic fusion and adaptive optimization. Full article

Distributed acoustic sensing (DAS) in pipeline safety warning systems confronts multiple challenges during technological evolution and application expansion, primarily including recognition accuracy, real-time performance, and the identification of weak signals for pipeline third-party intrusion (TPI) detection in complex environments. So, this paper proposes a Pipeline Fiber Optic Warning-Mixture of Experts (PFOW-MoE) method to address challenges in DAS systems. The proposed method is innovative in the sense that: (1) Multi-modal feature perception expert model design: Different intrusion behaviors are unique in the time, spatial, and frequency domains; (2) Efficient decision framework with dynamic gating mechanism: It evaluates input signal features in real time. (3) Robustness enhancement mechanism for weak signal perception: A weak signal detection branch is added to dynamic gating. Experimental validation on actual pipeline datasets shows PFOW-MoE achieves 98.27% accuracy on the entire sample set. On weak signal samples, it achieves 96.00%. The single-sample inference time is only 0.78 ms, meeting practical real-time engineering needs. Full article

To address challenges in complex group decision-making (GDM)—specifically preference fuzziness, intricate subgroup segmentation, and non-cooperative behavior—this study proposes an adaptive consensus model based on probabilistic linguistic term sets (PLTSs). By integrating fuzzy C-means (FCM) clustering with a Gaussian mixture model (GMM), a fuzzy Gaussian mixture model (FGMM) is constructed to achieve soft segmentation of expert preference distributions. On this basis, an adaptive consensus feedback mechanism is developed, which dynamically integrates interactive and automated adjustment strategies via multi-level consensus thresholds, thereby balancing decision efficiency and quality. To identify and control non-cooperative behaviors, a cooperation index and a three-tier management strategy, which incorporates intra-group negotiation, weight penalties and an exit-delegation mechanism, were introduced. In the case of strategic decision-making of new energy vehicles (NEV), after four rounds of feedback iterations, the group consensus level increased from the initial 0.316 to 0.804, reaching the preset threshold and verifying the effectiveness of the consensus mechanism. Compared with the existing literature methods, the framework in this paper achieves more comprehensive integration and innovation in four aspects: preference expression, clustering mechanism, consensus feedback and behavior management. Full article

Facial occlusion induces sample-wise reliability shifts in facial expression recognition (FER), where the usefulness of global context and local discriminative cues varies dramatically with the amount of visible facial information. Existing occlusion-robust FER studies often evaluate under limited or homogeneous occlusion settings and commonly adopt static fusion strategies, which are insufficient for complex and heterogeneous real-world occlusions. In this work, we establish a rigorous occlusion robustness evaluation protocol by constructing a fixed offline test benchmark with diverse synthetic occlusion patterns (e.g., masks, sunglasses, texture blocks, and mixed occlusions) on top of public FER test splits. We further propose a Dual-Stream Adaptive Weighting Mixture-of-Experts framework (DS-AW-MoE) that fuses a global contextual expert and a local discriminative expert via an occlusion-aware weighting network. Crucially, we introduce a facial visibility assessment as a task-agnostic prior to explicitly regulate expert contributions, enabling dynamic re-allocation of model capacity according to input-dependent feature reliability. Extensive experiments on public datasets and the constructed occlusion benchmark demonstrate that DS-AW-MoE achieves more stable recognition under complex occlusions, characterized by a smaller and more consistent performance drop. To support reproducibility under dataset license constraints, we will release an anonymous, fully runnable repository containing the complete occlusion synthesis pipeline, evaluation protocol, and configuration files, allowing researchers to reproduce the benchmark after obtaining the original datasets. Full article