Search Results (444)

Enhanced Geothermal Systems (EGS) extend geothermal energy beyond conventional hydrothermal resources but face challenges in creating sustainable heat exchangers in low-permeability formations. This review synthesizes achievements from the Utah Frontier Observatory for Research in Geothermal Energy (FORGE), a field laboratory advancing EGS readiness in 175–230 °C granitic basement. From 2017 to 2025, drilling, multi-stage hydraulic stimulation, and monitoring established feasibility and operating parameters for engineered reservoirs. Hydraulic connectivity was created between highly deviated wells with ~300 ft vertical separation via hydraulic and natural fracture networks, validated by sustained circulation tests achieving 10 bpm injection at 2–3 km depth. Advanced monitoring (DAS, DTS, and microseismic arrays) delivered fracture propagation diagnostics with ~1 m spatial resolution and temporal sampling up to 10 kHz. A data infrastructure of 300+ datasets (>133 TB) supports reproducible ML. Geomechanical analyses showed minimum horizontal stress gradients of 0.74–0.78 psi/ft and N–S to NNE–SSW fractures aligned with maximum horizontal stress. Near-wellbore tortuosity, driving treating pressures to 10,000 psi, underscores completion design optimization, improved proppant transport in high-temperature conditions, and coupled thermos-hydro-mechanical models for long-term prediction, supported by AI platforms including an offline Small Language Model trained on Utah FORGE datasets. Full article

This article analyzes the Solimene façade (Vietri sul Mare, Campania, Italy, 1952–1955). The survey, already acquired with active and passive sensors, was integrated with close-range photogrammetry of some sections of the main wall. The purpose of the new acquisitions was to generate data to inform a plug-in that, in the latest versions of the Revit software, correlates parametric and procedural environments. The focus of the study was the rationalization of the formal structure of the amphora, the heart of the main façade. Logic and geometric language guide the identification of a possible mathematical relationship aimed at parametrically modifying the model. The logical diagrams, converted into a Grasshopper preview, can be managed through graphical nodes. In the form of flowcharts (visual scripts), the finite sequence of procedural steps has the advantage of managing and modifying, in real time and in a user-friendly manner, the morphometric characteristics of the small “mummarella.” The results identify the morphometric characteristics common to a typological family composed of Vietri amphorae that, in the field of architectural design, uses the typical functions of system families. The goal is to approach sustainable and participatory design solutions by providing functions that can be graphically manipulated from within the software environment. Full article

The complex environment of electrical work sites presents hazards that are diverse in form, easily concealed, and difficult to distinguish from their surroundings. Due to poor model generalization, most traditional visual recognition methods are prone to errors and cannot meet the current safety management needs in electrical work. This paper presents a novel framework for hazard identification that integrates chain-of-thought reasoning and self-verification mechanisms within a visual-language large model (VLLM) to enhance accuracy. First, typical hazard scenario data for crane operation and escalator work areas were collected. The Janus-Pro VLLM model was selected as the base model for hazard identification. Then, designing a chain-of-thought enhanced the model’s capacity to identify critical information, including the status of crane stabilizers and the zones where personnel are located. Simultaneously, a self-verification module was designed. It leveraged the multimodal comprehension capabilities of the VLLM to self-check the identification results, outputting confidence scores and justifications to mitigate model hallucination. The experimental results show that integrating the self-verification method significantly improves hazard identification accuracy, with average increases of 2.55% in crane operations and 4.35% in escalator scenarios. Compared with YOLOv8s and D-FINE, the proposed framework achieves higher accuracy, reaching up to 96.3% in crane personnel intrusion detection, and a recall of 95.6%. It outperforms small models by 8.1–13.8% in key metrics without relying on massive labeled data, providing crucial technical support for power operation hazard identification. Full article

Machine learning inference is increasingly deployed on shared and cloud infrastructures, where both user inputs and model parameters are highly sensitive. Confidential computing promises to protect these assets using Trusted Execution Environments (TEEs), yet existing TEE-based inference systems remain fundamentally constrained: they rely almost exclusively on low-level, memory-unsafe languages to enforce confinement, sacrificing developer productivity, portability, and access to modern ML ecosystems. At the same time, mainstream high-level runtimes, such as Python, are widely considered incompatible with enclave execution due to their large memory footprints and unsafe model-loading mechanisms that permit arbitrary code execution. To bridge this gap, we present the first Python-based ML inference system that executes entirely inside Intel SGX enclaves while safely supporting untrusted third-party models. Our design enforces standardized, declarative model representations (ONNX), eliminating deserialization-time code execution and confining model behavior through interpreter-mediated execution. The entire inference pipeline (including model loading, execution, and I/O) remains enclave-resident, with cryptographic protection and integrity verification throughout. Our experimental results show that Python incurs modest overheads for small models (≈$17\%$) and outperforms a low-level baseline on larger workloads (97% vs. 265% overhead), demonstrating that enclave-resident high-level runtimes can achieve competitive performances. Overall, our findings indicate that Python-based TEE inference is practical and secure, enabling the deployment of untrusted models with strong confidentiality and integrity guarantees while maintaining developer productivity and ecosystem advantages. Full article

In the field of Natural Language Inference (NLI), model interpretability remains an urgent and unresolved challenge. Existing interpretability-oriented annotated datasets are highly limited, and manually constructing natural language explanations is both costly and inconsistent, making it difficult to balance model performance and interpretability. To address this issue, this paper proposes an interpretable NLI framework based on active learning, Explanation Generation Model-Prediction Model (EGM-PM), and designs an active learning sampling algorithm, Explanation-aware Transition from Clustering to Margin (ETCM), that incorporates natural-language explanation information. In this framework, Large Language Models (LLMs) are employed to automate explanation annotation, reducing dependence on human experts in traditional active learning. A small number of high-value samples obtained via ETCM sampling are used to train the EGM, whose generated natural-language explanations are then used to guide the PM in label inference. Experimental results show that data sampled by ETCM substantially enhance the model’s ability to learn relational and logical structures between premise–hypothesis pairs. Compared with other active learning algorithms, ETCM approaches full-data performance more rapidly while using significantly fewer labeled samples. This finding confirms the value of natural language explanation semantics in improving both model performance and interpretability. Furthermore, this paper employs prompt engineering to construct an interpretability-oriented NLI dataset, Explainable Natural Language Inference (ExNLI), which augments traditional premise–hypothesis pairs with natural-language explanations. Human and automated evaluations confirm the consistency and faithfulness of these explanations. The dataset has been publicly released, offering a low-cost and scalable data construction approach for future research on explainable NLI. Full article

The emergence of low-cost edge devices has enabled the integration of automatic speech recognition (ASR) into IoT environments, creating new opportunities for real-time language assessment. However, achieving reliable performance on resource-constrained hardware remains a significant challenge, especially on the Artificial Internet of Things (AIoT). This study presents an AIoT-based framework for automated English-speaking assessment that integrates architecture and system design, ASR benchmarking, and reliability analysis on edge devices. The proposed AIoT-oriented architecture incorporates a lightweight scoring framework capable of analyzing pronunciation, fluency, prosody, and CEFR-aligned speaking proficiency within an automated assessment system. Seven open-source ASR models—four Whisper variants (tiny, base, small, and medium) and three Vosk models—were systematically benchmarked in terms of recognition accuracy, inference latency, and computational efficiency. Experimental results indicate that Whisper-medium deployed on the Raspberry Pi 5 achieved the strongest overall performance, reducing inference latency by 42–48% compared with the Raspberry Pi 4 and attaining the lowest Word Error Rate (WER) of 6.8%. In contrast, smaller models such as Whisper-tiny, with a WER of 26.7%, exhibited two- to threefold higher scoring variability, demonstrating how recognition errors propagate into automated assessment reliability. System-level testing revealed that the Raspberry Pi 5 can sustain near real-time processing with approximately 58% CPU utilization and around 1.2 GB of memory, whereas the Raspberry Pi 4 frequently approaches practical operational limits under comparable workloads. Validation using real learner speech data (approximately 100 sessions) confirmed that the proposed system delivers accurate, portable, and privacy-preserving speaking assessment using low-power edge hardware. Overall, this work introduces a practical AIoT-based assessment framework, provides a comprehensive benchmark of open-source ASR models on edge platforms, and offers empirical insights into the trade-offs among recognition accuracy, inference latency, and scoring stability in edge-based ASR deployments. Full article

This systematic review examines 205 studies on the use of Large Language Models (LLMs) in psychiatry, psychology, psychotherapy, and clinical workflows. Furthermore, studies that directly evaluated at least one LLM in a mental health context were included in the extended detailed analysis. GPT-4 and GPT-3.5 were the most commonly assessed models. Although LLMs showed promising short-term performance across domains, most evaluations relied on small, non-longitudinal datasets and single-session testing, limiting generalizability. The evidence indicates rapid growth but significant methodological inconsistency, emphasizing the need for more diverse datasets, standardized evaluation, and long-term validation before clinical integration. This review also examines how LLMs are being incorporated into mental health practice, outlining key challenges, limitations, and emerging opportunities. Ethical, clinical, and technological considerations are proposed to guide responsible adoption. Given the complexity of mental health care, a multidisciplinary, human-centered approach remains essential to ensure that future LLM applications augment—rather than replace—professional expertise. Full article

Renewable energy sources like wind turbines and solar panels are integrated into modern power grids as Distributed Energy Resources (DERs). These DERs can operate independently or as part of microgrids. Interconnecting multiple microgrids creates Networked Microgrids (NMGs) that increase reliability, resilience, and independent power generation. However, the trustworthiness of individual DERs remains a critical challenge in NMGs, particularly when integrating previously deployed or geographically distributed units managed by entities with varying expertise. Assessing DER trustworthiness ensuring reliability and security is essential to prevent system-wide instability. Thisresearch addresses this challenge by proposing a lightweight trust metric generation system capable of processing structured and semi-structured DER data to produce key trust indicators. The system employs a Small Language Model (SLM) with approximately 16 million parameters for textual data understanding and metric extraction, followed by a regression head to output bounded trust scores. Designed for deployment in computationally constrained environments, the SLM requires only 64.6 MB of disk space and 200–250 MB of memory that is significantly lesser than larger models such as DeepSeek R1, Gemma-2, and Phi-3, which demand 3–12 GB. Experimental results demonstrate that the SLM achieves high correlation and low mean error across all trust metrics while outperforming larger models in efficiency. When integrated into a full neural network-based trust framework, the generated metrics enable accurate prediction of DER trustworthiness. These findings highlight the potential of lightweight SLMs for reliable and resource-efficient trust assessment in NMGs, supporting resilient and sustainable energy systems in smart cities. Full article

Background/Objective: The primary cilium is the sensory organelle of a cell and a dynamic membrane protrusion during the cell cycle. It originates from the centriole at G₀/G₁ and undergoes disassembly to release centrioles for spindle formation before a cell enters mitosis, thereby serving as a cell cycle checkpoint. Cancer cells that undergo rapid cell cycle and replication have a low ciliation rate. In this study, we aimed to identify cilia-promoting drugs that can accelerate ciliation and decelerate replication of cancer cells. Methods: To perform a comprehensive and efficient literature search on drugs that can promote ciliation, we developed an intelligent process that integrates either the GPT 4 Turbo, Gemini 1.5 Pro, or Claude 3.5 Haiku application programming interfaces (APIs) into a PubMed scraper that we coded, enabling the large language models (LLMs) to directly query articles for predefined user questions. We evaluated the performance of this intelligent literature search based on metrics and tested the effect of two candidate drugs on ciliation and proliferation of medulloblastoma cells. Results: Gemini was the best model overall, as it balanced high accuracy with solid precision and recall scores. Among the top candidate drugs identified are Alvocidib and Alisertib, small-molecule inhibitors of CILK1 and AURKA, respectively. Here, we show that both kinase inhibitors can effectively increase cilia frequency and significantly decrease the replication of medulloblastoma cells. Conclusions: The results demonstrated the potential of using cilia-promoting drugs, such as Alvocidib and Alisertib, to suppress cancer cell replication. Additionally, it shows the massive benefits of integrating accessible large language models to conduct sweeping, rapid, and accurate literature searches. Full article

Background/Objectives: Interpretation of chest X-rays (CXRs) requires accurate identification of lesion presence, diagnosis, location, size, and number to be considered complete. However, the effectiveness of large language models with vision capabilities (LLMs) in performing these tasks remains uncertain. This study aimed to evaluate the image-only interpretation performance of LLMs in the absence of clinical information. Methods: A total of 247 CXRs covering 13 diagnostic categories, including pulmonary edema, cardiomegaly, lobar pneumonia, and other conditions, were evaluated using Gemini 1.0, Gemini 1.5 Pro, GPT-4 Turbo, and GPT-4o. The text outputs generated by the LLMs were evaluated at two levels: (1) primary diagnosis accuracy across the 13 predefined diagnostic categories, and (2) identification of key imaging features described in the generated text. Primary diagnosis accuracy was assessed based on whether the model correctly identified the target diagnostic category and was classified as fully correct, partially correct, or incorrect according to predefined clinical criteria. Non-diagnostic imaging features, such as posteroanterior and anteroposterior (PA/AP) views, side markers, foreign bodies, and devices, were recorded and analyzed separately rather than being incorporated into the primary diagnostic scoring. Results: When fully and partially correct responses were treated as successful detections, vLLMs showed higher sensitivity for large, bilateral, multiple lesions and prominent devices, including acute pulmonary edema, lobar pneumonia, multiple malignancies, massive pleural effusions, and pacemakers, all of which demonstrated statistically significant differences across categories in chi-square analyses. Feature descriptions varied among models, especially in PA/AP views and side markers, though central lines were partially recognized. Across the entire dataset, Gemini 1.5 Pro achieved the highest overall detection rate, followed by Gemini 1.0, GPT-4o, and GPT-4 Turbo. Conclusions: Although LLMs were able to identify certain diagnoses and key imaging features, their limitations in detecting small lesions, recognizing laterality, reasoning through differential diagnoses, and using domain-specific expressions indicate that CXR interpretation without textual cues still requires further improvement. Full article

Large Language Models (LLMs) are vulnerable to Indirect Prompt Injection Attacks (IPIAs), where malicious instructions are embedded within external content rather than direct user input. This study presents an embedding-based detection approach that analyses the semantic relationship between user intent and external content, enabling the early identification of IPIAs that conventional defences overlook. We also provide a dataset of 70,000 samples, constructed using 35,000 malicious instances from the Benchmark for Indirect Prompt Injection Attacks (BIPIA) and 35,000 benign instances generated using ChatGPT-4o-mini. Furthermore, we performed a comparative analysis of three embedding models, namely OpenAI text-embedding-3-small, GTE-large, and MiniLM-L6-v2, evaluated in combination with XGBoost, LightGBM, and Random Forest classifiers. The best-performing configuration using OpenAI embeddings with XGBoost achieved an accuracy of 97.7% and an F1-score of 0.977, matching or exceeding the performance of existing IPIA detection methods while offering practical deployment advantages. Unlike prevention-focused approaches that require modifications to the underlying LLM architecture, the proposed method operates as a model-agnostic external detection layer with an average inference time of 0.001 ms per sample. This detection-based approach complements existing prevention mechanisms by providing a lightweight, scalable solution that can be integrated into LLM pipelines without requiring architectural changes. Full article

Massive MIMO systems can substantially enhance spectral efficiency, but such gains rely on the availability of accurate channel state information (CSI). However, the increase in the number of antennas leads to a significant growth in feedback overhead, while conventional deep-learning-based CSI feedback methods also impose a substantial computational burden on the user equipment (UE). To address these challenges, this paper proposes LLM4FB, a one-sided CSI feedback framework that leverages a pre-trained large language model (LLM). In this framework, the UE performs only low-complexity linear projections to compress CSI. In contrast, the BS leverages a pre-trained LLM to accurately reconstruct and predict CSI. By utilizing the powerful modeling capabilities of the pre-trained LLM, only a small portion of the parameters needs to be fine-tuned to improve CSI recovery accuracy with low training cost. Furthermore, a multiobjective loss function is designed to simultaneously optimize normalized mean square error (NMSE) and spectral efficiency (SE). Simulation results show that LLM4FB outperforms existing methods across various compression ratios and mobility levels, achieving high-precision CSI feedback with minimal computational capability from terminal devices. Therefore, LLM4FB presents a highly promising solution for next-generation wireless sensor networks and industrial IoT applications, where terminal devices are often strictly constrained by energy and hardware resources. Full article

Cyberbullying and hate speech increasingly appear in multi-modal social media posts, where images and text are combined in diverse and fast changing ways across platforms. These posts differ in style, vocabulary and layout, and labeled data are sparse and noisy, which makes it difficult to train detectors that are both reliable and deployable under tight computational budgets. Many high performing systems rely on large vision language backbones, full parameter fine tuning, online retrieval or model ensembles, which raises training and inference costs. We present a parameter efficient cross-platform multi-modal transfer learning framework for cyberbullying and hateful content detection. Our framework has three components. First, we perform domain adaptive pretraining of a compact ViLT backbone on in domain image-text corpora. Second, we apply parameter efficient fine tuning that updates only bias terms, a small subset of LayerNorm parameters and the classification head, leaving the inference computation graph unchanged. Third, we use noise aware knowledge distillation from a stronger teacher built from pretrained text and CLIP based image-text encoders, where only high confidence, temperature scaled predictions are used as soft labels during training, and teacher models and any retrieval components are used only offline. We evaluate primarily on Hateful Memes and use IMDB as an auxiliary text only benchmark to show that the deployment aware PEFT + offline-KD recipe can still be applied when other modalities are unavailable. On Hateful Memes, our student updates only 0.11% of parameters and retain about 96% of the AUROC of full fine-tuning. Full article

Entity alignment (EA) facilitates knowledge fusion by matching semantically identical entities in distinct knowledge graphs (KGs). Existing embedding-based methods rely solely on intrinsic KG facts and often struggle with long-tail entities due to insufficient information. Recently, large language models (LLMs), empowered by rich background knowledge and strong reasoning abilities, have shown promise for EA. However, most current LLM-enhanced approaches follow the in-context learning paradigm, requiring multi-round interactions with carefully designed prompts to perform additional auxiliary operations, which leads to substantial computational overhead. Moreover, they fail to fully exploit the complementary strengths of embedding-based small models and LLMs. To address these limitations, we propose HIEA, a novel hierarchical inference framework for entity alignment. By instruction-tuning a generative LLM with a unified and concise prompt and a knowledge adapter, HIEA produces alignment results with a single LLM invocation. Meanwhile, embedding-based small models not only generate candidate entities but also support the LLM through data augmentation and certainty-aware source entity classification, fostering deeper collaboration between small models and LLMs. Extensive experiments on both standard and highly heterogeneous benchmarks demonstrate that HIEA consistently outperforms existing embedding-based and LLM-enhanced methods, achieving absolute Hits@1 improvements of up to 5.6%, while significantly reducing inference cost. Full article

This paper presents a scalable and low-cost Retrieval Augmented Generation (RAG) architecture designed to enhance learning in university-level courses, with a particular focus on supporting students from economically disadvantaged backgrounds. Recent advances in large language models (LLMs) have demonstrated considerable potential in educational contexts; however, their adoption is often limited by computational costs and the need for stable broadband access, issues that disproportionately affect low-income learners. To address this challenge, we propose a lightweight, mobile, and friendly RAG system that integrates the LLaMA language model with the Milvus vector database, enabling efficient on device retrieval and context-grounded generation using only modest hardware resources. The system was implemented in a university-level Data Mining course and evaluated over four semesters using a quasi-experimental design with randomized assignment to experimental and control groups. Students in the experimental group had voluntary access to the RAG assistant, while the control group followed the same instructional schedule without exposure to the tool. The results show statistically significant improvements in academic performance for the experimental group, with p < 0.01 in the first semester and p < 0.001 in the subsequent three semesters. Effect sizes, measured using Hedges g to account for small cohort sizes, increased from 0.56 (moderate) to 1.52 (extremely large), demonstrating a clear and growing pedagogical impact over time. Qualitative feedback further indicates increased learner autonomy, confidence, and engagement. These findings highlight the potential of mobile RAG architectures to deliver equitable, high-quality AI support to students regardless of socioeconomic status. The proposed solution offers a practical engineering pathway for institutions seeking inclusive, scalable, and resource-efficient approaches to AI-enhanced education. Full article