Papers

This research presents a comprehensive diagnostic study of confidence-gated iterative induction for zero-shot Named Entity Recognition (NER) in crisis scenarios. While existing approaches struggle to adapt to novel disaster lexicons without manually curated resources, we investigate whether iterative knowledge distillation can overcome these limitations. Our framework leverages a pretrained language model to extract high-recall entity candidates, then iteratively distills domain knowledge through a self-correcting loop that uses high-confidence seeds to induce micro-gazetteers and syntactic rules. Comprehensive evaluations on synthetic crisis data reveal that the framework maintains a constant zero-shot F1-score of approximately 0.295 across all experimental configurations, demonstrating that the iterative mechanism provides no measurable improvement over baseline approaches. This negative result offers valuable diagnostic insights into the fundamental challenges of adaptive NER in dynamic crisis domains, including confidence threshold calibration difficulties, clustering algorithm limitations, and error propagation risks. The findings provide a cautionary tale for researchers working on adaptive NER systems and establish a foundation for future research on more robust zero-shot approaches in crisis scenarios.

Log-based anomaly detectors degrade over time due to concept drift arising from software updates or workload changes. Existing systems typically react by retraining entire models, leading to catastrophic forgetting and inefficiencies. We propose an adaptive framework that first classifies drift in log data into semantic (frequency shifts within known templates) and syntactic (emergence of new log templates) categories via statistical tests and novelty detection. Based on the identified drift type, a policy-driven lifelong learning manager applies targeted updates---experience replay to mitigate forgetting under semantic drift and dynamic model expansion to accommodate syntactic drift. This approach is validated on semi-synthetic logs and real-world longitudinal datasets (HDFS, Apache, and BGL), maintaining high F1-scores, reducing computational overhead, and preserving historical knowledge compared to monolithic retraining.

Financial text extraction faces serious challenges in multi-entity sentiment attribution and numerical sensitivity, often leading to pitfalls in real-world deployment. In this work, we propose ConFIT (Contrastive Financial Information Tuning), a knowledge-guided contrastive learning framework that employs a Semantic-Preserving Perturbation (SPP) engine to generate high-quality, programmatically synthesized hard negatives. By integrating domain knowledge sources such as the Loughran-McDonald lexicon and Wikidata, and applying rigorous perplexity and Natural Language Inference (NLI) filtering, ConFIT trains language models to differentiate subtle perturbations in financial statements. Evaluations on FiQA and SENTiVENT using FinBERT and Llama-3 8B show both promise improvements and unexpected pitfalls, highlighting challenges that warrant further research.

Industrial fault detection systems often struggle to distinguish benign operational drifts (e.g., tool wear, recipe changes) from incipient faults, frequently adapting to faults as new ``normal'' states and risking catastrophic failures. This work proposes a hierarchical framework that decouples change detection from change characterization. When a drift is detected, the system generates a Multi-Scale Change Signature (MSCS) that quantifies geometric and statistical transformations in the primary detector’s latent space. An unsupervised Drift Characterization Module (DCM), trained on an Online Normality Baseline (ONB), classifies each signature as benign or potentially faulty. Benign drifts are ignored, while potential faults are flagged for review; confirmed benign drifts are incorporated into the ONB for future adaptation. The framework is model-agnostic, computationally efficient, and scalable through a tiered human-in-the-loop mechanism. Experiments on the Tennessee Eastman Process dataset with injected drifts and faults demonstrate high fault detection rates, fewer false alarms, and efficient adaptation to benign changes.

Wearable Human Activity Recognition (HAR) systems face significant performance degradation when sensors are placed at different body locations or orientations. We introduce a hierarchical adaptive normalization method that addresses these challenges through a two-stage cascade. The first stage combines gravity-based orientation correction with placement context inference using signal variance analysis, while a novel stability gate prevents harmful adaptation during unstable periods. The second stage employs placement-conditioned adaptive Batch Normalization to refine feature representations in real-time. Comprehensive evaluations on public and custom datasets show that our method achieves 0.847±0.023 macro F1-score, outperforming static baselines by 36\% and state-of-the-art unsupervised domain adaptation methods by 13.7\%. The approach maintains real-time performance with only 2.3ms inference time and 45.2MB memory usage, demonstrating practical viability for on-device deployment in dynamic real-world scenarios.

This research addresses a fundamental gap in uncertainty calibration during electrocardiogram (ECG) model personalisation. We propose \emph{Adaptive Evidential Meta-Learning}, a framework that attaches a lightweight evidential head with hyper-network-conditioned priors to a frozen ECG foundation model. The hyper-network dynamically sets the evidential prior using robust, class-conditional statistics computed from a few patient-specific ECG samples. Trained via a two-stage meta-curriculum, our approach enables rapid adaptation with well-calibrated uncertainty estimates, making it highly applicable for real-world clinical deployment where both prediction accuracy and uncertainty awareness are crucial.

The rapid deployment of 5G networks has intensified concerns about energy consumption in mobile communication systems. Unlike previous generations, 5G base stations (BSs) exhibit significant power draw even under zero traffic conditions, with static power accounting for $30\sim 40\%$ of total energy consumption. This paper proposes a novel data-driven framework that decouples total base station energy consumption into static and dynamic components, enabling more precise energy optimization. For static consumption modeling, we introduce a hybrid ResNet-XGBoost architecture that processes configuration parameters including bandwidth, antenna elements, transmit power, carrier count, and tilt angle. For dynamic consumption, we implement a Tabular Probabilistic Function Network (TabPFN) to capture the nonlinear relationship between resource utilization and energy demand. Experimental results using real-world data from a provincial Chinese telecom operator demonstrate that our model achieves a $15.5\%$ reduction in Mean Absolute Error (MAE) and an $R^2$ of 0.91 compared to conventional approaches.

Knowledge Graphs (KGs) have long served as a fundamental infrastructure for structured knowledge representation and reasoning. With the advent of Large Language Models (LLMs), the construction of KGs has entered a new paradigm—shifting from rule-based and statistical pipelines to language-driven and generative frameworks. This survey provides a comprehensive overview of recent progress in **LLM-empowered knowledge graph construction**, systematically analyzing how LLMs reshape the classical three-layered pipeline of ontology engineering, knowledge extraction, and knowledge fusion. We first revisit traditional KG methodologies to establish conceptual foundations, and then review emerging LLM-driven approaches from two complementary perspectives: *schema-based* paradigms, which emphasize structure, normalization, and consistency; and *schema-free* paradigms, which highlight flexibility, adaptability, and open discovery. Across each stage, we synthesize representative frameworks, analyze their technical mechanisms, and identify their limitations. Finally, the survey outlines key trends and future research directions, including KG-based reasoning for LLMs, dynamic knowledge memory for agentic systems, and multimodal KG construction. Through this systematic review, we aim to clarify the evolving interplay between LLMs and knowledge graphs, bridging symbolic knowledge engineering and neural semantic understanding toward the development of adaptive, explainable, and intelligent knowledge systems.

Artificial Intelligence (AI) has great potential to transform rock mechanics by tackling its inherent complexities, such as anisotropy, nonlinearity, discontinuous, and multiphase nature. This review explores the evolution of AI, from basic neural networks like the BP model to advanced architectures such as Transformers, and their applications in areas like microstructure reconstruction, prediction of mechanical parameters, and addressing engineering challenges such as rockburst prediction and tunnel deformation. Machine learning techniques, particularly Convolutional Neural Networks (CNNs) and Generative Adversarial Networks (GANs), have been crucial in automating tasks like fracture detection and efficiently generating 3D digital rock models. However, the effectiveness of AI in rock mechanics is limited by data scarcity and the need for high-quality datasets. Hybrid approaches, such as combining physics-informed neural networks (PINNs) with traditional numerical methods, offer promising solutions for solving governing equations. Additionally, Large Language Models (LLMs) are emerging as valuable tools for code generation and decision-making support. Despite these advancements, challenges remain, including issues with reproducibility, model interpretability, and adapting AI models to specific domains. Future progress will hinge on the availability of improved datasets, greater interdisciplinary collaboration, and the integration of spatial intelligence frameworks to bridge the gap between AI’s theoretical potential and its practical application in rock engineering.

Intelligent rock mechanics represents the convergence of artificial intelligence (AI) and classical rock mechanics, providing new paradigms to understand, model, and predict the complex behaviors of geological materials. This review synthesizes recent progress from foundational AI methodologies to their practical applications in rock engineering. Traditional challenges—such as anisotropy, discontinuities, and multiphysics coupling—have been re-examined through data-driven and hybrid approaches that integrate learning algorithms with physical principles. The study traces the evolution of AI in this field, from early backpropagation and support vector machines to modern deep learning frameworks such as convolutional and transformer architectures, highlighting their roles in microstructure reconstruction, mechanical parameter estimation, constitutive modeling, and real-time hazard prediction. Emerging techniques, including physics-informed neural networks and graph-based learning, bridge data-driven inference with physical interpretability, while large language models are beginning to facilitate automated code generation and decision support in geotechnical analysis. Despite remarkable progress, key challenges remain in data quality, model generalization, and interpretability. Addressing these issues requires standardized datasets, interdisciplinary collaboration, and the establishment of transparent, reproducible AI workflows. The paper concludes by outlining a forward-looking perspective on developing next-generation intelligent frameworks capable of coupling physical knowledge, spatial reasoning, and adaptive learning, thereby advancing rock mechanics from empirical modeling toward fully intelligent, autonomous systems.

This paper proposes a federated model of AI Scientists, integrating a layered stack architecture, an iterative discovery pipeline, and a governance-aligned roadmap. We argue that AI Scientists should not only accelerate discovery but also serve as custodians of epistemic integrity. Through case studies in drug discovery, climate modeling, and materials science, we demonstrate how federation enables cross-domain synthesis while embedding reproducibility, incentive alignment, and participatory governance. We conclude with a research roadmap toward Trusted AI Scientists, highlighting technical, incentive, and governance challenges.

Growing demands for clinical data privacy and storage constraints have spurred advances in Source Free Unsupervised Domain Adaptation (SFUDA). SFUDA addresses the domain shift by adapting models from the source domain to the unseen target domain without accessing source data, even when target-domain labels are unavailable. However, SFUDA faces significant challenges: the absence of source domain data and label supervision in the target domain due to source free and unsupervised settings. To address these issues, we propose HEAL, a novel SFUDA framework that integrates Hierarchical denoising, Edge-guided selection, size-Aware fusion, and Learning-free characteristic. Large-scale cross-modality experiments demonstrate that our method outperforms existing SFUDA approaches, achieving state-of-the-art (SOTA) performance. The source code is publicly available at: https://anonymous.4open.science/r/HEAL-10C5.

Training small language models is challenging due to their limited capacity to capture complex patterns and their susceptibility to overfitting. To address these issues, we investigate gradient noise injection as a regularization strategy, building on prior work while introducing a noise schedule that decays exponentially over training. Unlike existing techniques, our method explicitly controls the trade-off between exploration and stability during optimization. We compare the exponential decay schedule with linear and adaptive variants, demonstrating empirically that the exponential schedule yields superior convergence and generalization. Extensive experiments on diverse text corpora, including shakespeare\_char, enwik8, text8, and larger benchmark datasets, show consistent improvements in training dynamics, validation loss, and final performance. We report error bars and statistical significance tests to ensure robustness of the results. Detailed implementation information, including model architectures, hyperparameter settings, dataset sizes, and optimization strategies, is provided to support reproducibility, and we release our code and trained models publicly. Furthermore, we compare gradient noise injection with other regularization methods such as dropout, weight decay, and data augmentation, both in isolation and in combination, revealing complementary effects on training stability and generalization. Finally, we analyze the computational cost of gradient noise injection relative to these baselines, highlighting its practical efficiency in resource-constrained environments. Together, these contributions position gradient noise injection as a theoretically grounded, empirically validated, and computationally practical method for improving the robustness of small language models.

Large language models (LLMs) struggle to effectively utilize a growing number of external tools, such as those defined by the Model Context Protocol (MCP)[ 1], due to prompt bloat and selection complexity. We introduce RAG-MCP, a Retrieval-Augmented Generation framework that overcomes this challenge by offloading tool discovery. RAGMCP uses semantic retrieval to identify the most relevant MCP(s) for a given query from an external index before engaging the LLM. Only the selected tool descriptions are passed to the model, drastically reducing prompt size and simplifying decision-making. Experiments, including an MCP stress test, demonstrate RAG-MCP significantly cuts prompt tokens (e.g., by over 50%) and more than triples tool selection accuracy (43.13% vs 13.62% baseline) on benchmark tasks. RAG-MCP enables scalable and accurate tool integration for LLMs.