Search Results (53)

The proliferation of intelligent sensor networks in urban surveillance and remote sensing has triggered the explosive growth of unstructured visual sensor data. Accurately retrieving targets from these massive streams based on complex cross-modal user intents remains a critical bottleneck for efficient intelligent perception. Composed Image Retrieval (CIR) addresses this by enabling retrieval via a multi-modal query that combines a reference image with semantic control signals. However, existing methods often struggle with abstract instructions in real-world scenarios. Consequently, models often suffer from feature distribution shifts due to focus ambiguity, as well as semantic erosion caused by highly entangled visual and textual features. To address these challenges, we propose a geometry-based Selective Orthogonal Projection Network (SOP). First, the Selective Focus Recovery module quantifies instruction uncertainty via information entropy and calibrates shifted query features to the true target distribution using structural consistency regularization. Second, to ensure data fidelity, we introduce Orthogonal Subspace Projectionand Geometric Composition Fidelity. These mechanisms employ Gram–Schmidt orthogonalization to decouple features into a constant visual base and an orthogonal modification increment, restricting semantic modifications to the null space. Extensive experiments on FashionIQ, Shoes, and CIRR datasets demonstrate that SOP significantly outperforms SOTA methods, offering a novel solution for efficient large-scale sensor data retrieval and analysis. Full article

Analyzing fine-grained student actions across institutions can drive timely feedback, early warning, and personalized support, yet it is constrained by privacy regulations, heterogeneous curricula, and non-IID behavior logs. This paper introduces BAF–FedLLM, a behavior-aware federated modeling framework that adapts large language models to next-action and outcome prediction without centralizing student data. The key idea is to treat multichannel interaction streams as semantically typed action tokens linked by a learned ActionGraph, and to align their temporal structure with an LLM through behavior prompts that inject domain context (task, resource, pedagogy, and affordance cues). We propose three novel components: (i) BP–FIT, a behavior-prompted federated instruction tuning scheme that trains low-rank adapters locally and aggregates them with secure masking and Rényi–DP accounting to ensure client-level privacy; (ii) ProtoAlign, a cross-client prototype contrastive objective that shares only noisy class-conditional anchors via secure aggregation to mitigate drift under non-IID partitions; and (iii) CBR, a causal behavior regularizer that penalizes intervention-sensitive shortcuts by enforcing invariance of predicted risks across detected instructional regimes. We further derive convergence guarantees for federated instruction tuning with noisy, partial participation and provide end-to-end privacy bounds. On three public education datasets (EdNet, ASSISTments, and OULAD) with institution-level partitions, BAF–FedLLM improves next-action AUC by 4.2–7.1% over strong federated baselines while reducing expected calibration error by up to 28% and communication by $5\times$ through adapter sparsity, under a typical privacy budget of $\epsilon \approx 1.7$ at $\delta ={10}^{-5}$. These results indicate that behavior-aware prompting and prototype alignment make LLMs practical for privacy-preserving student action analysis at scale, offering a principled path to deployable, regulation-compliant analytics across diverse learning ecosystems. Full article

Recent advancements in Artificial Intelligence and eXtended Reality (XR) have laid solid foundations for the development of a new paradigm in industrial maintenance under the light of Industry 5.0 framework. This research presents the design, development, and implementation of an XR-enabled remote maintenance framework that integrates real-time video collaboration, AI-assisted guidance, and a persistent digital asset knowledge layer based on Asset Administration Shells for Maintenance and Repair Operations (MRO). By combining fine-tuned Large Language Models (LLMs) with immersive XR interfaces, the proposed framework enables technicians to interact with virtual representations of industrial assets, access contextual instructions, and receive expert support remotely in real-time. Through seamless integration of historical MRO data, digital twins, and real-time sensor streams, the system facilitates dynamic fault diagnostics and Remaining Useful Life (RUL) estimation. Therefore, the proposed approach is positioned as a Metaverse-aligned implementation, combining synchronous multi-user collaboration, digital–physical coupling through digital twins, and semantic interoperability. The framework is validated through two industrial case studies, demonstrating its feasibility and practical impact on maintenance efficiency and knowledge transfer. The findings position the Industrial Metaverse as a transformative enabler in the future of AI-driven machinery health monitoring. Full article

As one of the most widely used high-performance kernels, General Matrix Multiplication, or GEMM, plays a pivotal role in diverse application fields. With the growing prevalence of training for Convolutional Neural Networks (CNNs) and Large Language Models (LLMs), the design and implementation of high-efficiency, low-precision GEMM on modern Neural Processing Unit (NPU) platforms are of great significance. In this work, HGEMM for Ascend NPU is presented, which enables collaborative processing of different computation types by Cube units and Vector units. The major contributions of this work are the following: (i) dual-stream pipeline scheduling is implemented, which synchronizes padding operations, matrix–matrix multiplications, and element-wise instructions across hierarchical buffers and compute units; (ii) a suite of tiling strategies and a corresponding strategy selection mechanism are developed, comprehensively accounting for the impacts from M, N, and K directions; and (iii) SplitK as well as ShuffleK methods are raised to address the challenges of memory access efficiency and AI Core utilization. Extensive evaluations demonstrate that our proposed HGEMM achieves an average 3.56× speedup over the CATLASS template-based implementation under identical Ascend NPU configurations, and an average 2.10× speedup relative to the cuBLAS implementation on Nvidia A800 GPUs under general random workloads. It also achieves a maximum computational utilization exceeding 90% under benchmark workloads. Moreover, the proposed HGEMM not only significantly outperforms the CATLASS template-based implementation but also delivers efficiency comparable to the cuBLAS implementation in OPT-based bandwidth-limited LLM inference workloads. Full article

Posture, defined as the body’s alignment relative to gravity, plays a vital role in musculoskeletal health by influencing muscle efficiency, joint integrity, and overall balance. The global shift to remote and sedentary work environments during the COVID-19 pandemic has amplified concerns regarding posture-related disorders and long-term ergonomic risks. This study introduces ALIGN, an IoT-enabled intelligent system for real-time sitting posture detection that integrates both machine learning and deep learning methodologies. Implemented on a single-board computer, the system processes live video streams to classify user posture as correct or incorrect and provides alert-based notifications when sustained improper posture is detected, thereby supporting real-time posture awareness without issuing corrective instructions. Among conventional classifiers, K-Nearest Neighbors (KNN), Support Vector Classifiers (SVC), and Multi-Layer Perceptrons (MLP) achieved accuracies of 98.74%, 96.64%, and 97.17%, respectively, while in the deep learning category, ResNet52 reached a test accuracy of 94.37%, outperforming DenseNet121 (81.53%). By enabling intelligent real-time detection and monitoring, ALIGN offers a scalable and cost-effective solution for ergonomic risk awareness and preventive digital health support. Full article

As artificial intelligence (AI) becomes increasingly woven into educational design and decision-making, its use within initial teacher education (ITE) exposes deep tensions between efficiency, equity, and professional agency. A critical action research study conducted across three iterations of a third-year ITE course investigated how pre-service teachers engaged with AI-supported lesson planning tools while learning to design for inclusion. Analysis of 123 lesson plans, reflective journals, and survey data revealed a striking pattern. Despite instruction in inclusive pedagogy, most participants reproduced fixed-tiered differentiation and deficit-based assumptions about learners’ abilities, a process conceptualised as micro-streaming. AI-generated recommendations often shaped these outcomes, subtly reinforcing hierarchies of capability under the guise of personalisation. Yet, through iterative reflection, dialogue, and critical framing, participants began to recognise and resist these influences, reframing differentiation as design for diversity rather than classification. The findings highlight the paradoxical role of AI in teacher education, as both an amplifier of inequity and a catalyst for critical consciousness and argue for the urgent integration of critical digital pedagogy within ITE programmes. AI can advance inclusive teaching only when educators are empowered to interrogate its epistemologies, question its biases, and reclaim professional judgement as the foundation of ethical pedagogy. Full article

Educational videos contain long periods of visual redundancy, where only a few frames convey meaningful instructional information. Conventional video models, which are designed for dynamic scenes, often fail to capture these subtle pedagogical transitions. We introduce LEARNet, an entropy-aware framework that models educational video understanding as the extraction of high-information instructional content from low-entropy visual streams. LEARNet combines a Temporal Information Bottleneck (TIB) for selecting pedagogically significant keyframes with a Spatial–Semantic Decoder (SSD) that produces fine-grained annotations refined through a proposed Relational Consistency Verification Network (RCVN). This architecture enables the construction of EVUD-2M, a large-scale benchmark with multi-level semantic labels for diverse instructional formats. LEARNet achieves substantial redundancy reduction (70.2%) while maintaining high annotation fidelity (F1 = 0.89, mAP@50 = 0.88). Grounded in information-theoretic principles, LEARNet provides a scalable foundation for tasks such as lecture indexing, visual content summarization, and multimodal learning analytics. Full article

Most adaptive learning systems personalize around content sequencing and difficulty adjustment rather than transforming instructional material within the lesson itself. This paper presents the STREAM (Semantic Transformation and Real-Time Educational Adaptation Multimodal) framework. This modular pipeline decomposes multimodal educational content into semantically tagged, pedagogically annotated units for regeneration into alternative formats while preserving source traceability. STREAM is designed to integrate automatic speech recognition, transformer-based natural language processing, and planned computer vision components to extract instructional elements from teacher explanations, slides, and embedded media. Each unit receives metadata, including time codes, instructional type, cognitive demand, and prerequisite concepts, designed to enable format-specific regeneration with explicit provenance links. For a predefined visual-learner profile, the system generates annotated path diagrams, two-panel instructional guides, and entity pictograms with complete back-link coverage. Ablation studies confirm that individual components contribute measurably to output completeness without compromising traceability. This paper reports results from a tightly scoped feasibility pilot that processes a single five-minute elementary STEM video offline under clean audio–visual conditions. We position the pilot’s limitations as testable hypotheses that require validation across diverse content domains, authentic deployments with ambient noise and bandwidth constraints, multiple learner profiles, including multilingual students and learners with disabilities, and controlled comprehension studies. The contribution is a transparent technical demonstration of feasibility and a methodological scaffold for investigating whether within-lesson content transformation can support personalized learning at scale. Full article

Vision–Language Models (VLMs) have become key contributors to the state of the art in contextual emotion recognition, demonstrating a superior ability to understand the relationship between context, facial expressions, and interactions in images compared to traditional approaches. However, their reliance on contextual cues can introduce unintended biases, especially when the background does not align with the individual’s true emotional state. This raises concerns for the reliability of such models in real-world applications, where robustness and fairness are critical. In this work, we explore the limitations of current VLMs in emotionally ambiguous scenarios and propose a method to overcome contextual bias. Existing VLM-based captioning solutions tend to overweight background and contextual information when determining emotion, often at the expense of the individual’s actual expression. To study this phenomenon, we created synthetic datasets by automatically extracting people from the original images using YOLOv8 and placing them on randomly selected backgrounds from the Landscape Pictures dataset. This allowed us to reduce the correlation between emotional expression and background context while preserving body pose. Through discriminative analysis of VLM behavior on images with both correct and mismatched backgrounds, we find that in 93% of the cases, the predicted emotions vary based on the background—even when models are explicitly instructed to focus on the person. To address this, we propose a multimodal approach (named BECKI) that incorporates body pose, full image context, and a novel description stream focused exclusively on identifying the emotional discrepancy between the individual and the background. Our primary contribution is not just in identifying the weaknesses of existing VLMs, but in proposing a more robust and context-resilient solution. Our method achieves up to 96% accuracy, highlighting its effectiveness in mitigating contextual bias. Full article

This study examines how different instructional media—face-to-face classes, live streaming, pre-recorded videos, and educational games—affect student learning outcomes in finance education. A sample of first-year economics students was assessed on their knowledge of basic financial principles before being randomly assigned to five groups. Four groups attended the same finance course delivered through different media formats, while a fifth group served as a control and received no instruction. After the course, all students completed a second (post-course) assessment. By comparing individual pre- and post-test results, as well as learning gains across the groups, we evaluated the effectiveness of each delivery method. The results show that all four instructional formats significantly improved financial knowledge compared to the control group. Among the media types, educational games proved to be an effective and reliable tool for delivering finance content. However, the differences in learning gains between face-to-face instruction, live streaming, and pre-recorded videos were not statistically significant. These findings indicate that a range of delivery models can be used effectively in finance education. The study contributes to current debates on cost-effective teaching strategies and supports evidence-based decisions on curriculum design in digitally transformed higher education environments after COVID-19. Full article

Accurate dynamic hydrological models are often too complex and costly for the rapid, broad-scale screening necessitated for proactive land-use planning against increasing cloudburst risks. This paper demonstrates the use of GPT-4 to develop a GUI-based Python 3.13.2 application for geospatial flood risk assessments. The study used instructive prompt techniques to script a traditional stream and catchment delineation methodology, further embedding it with a custom GUI. The resulting application demonstrates high performance, processing a 29.63 km² catchment at a 1 m resolution in 30.31 s, and successfully identifying the main upstream contributing areas and flow paths for a specified area of interest. While its accuracy is limited by terrain data artifacts causing stream breaks, this study demonstrates how human–AI collaboration, with the LLM acting as a coding assistant guided by domain expertise, can empower domain experts and facilitate the development of advanced GIS-based decision-support systems. Full article

This paper presents a new matrix representation of ant colony optimization (ACO) for solving parametric problems. This representation allows us to perform calculations using matrix processors and single-instruction multiple-data (SIMD) calculators. To solve the problem of stagnation of the method without a priori information about the system, a new probabilistic formula for choosing the parameter value is proposed, based on the additive convolution of the number of pheromone weights and the number of visits to the vertex. The method can be performed as parallel calculations, which accelerates the process of determining the solution. However, the high speed of determining the solution should be correlated with the high speed of calculating the objective function, which can be difficult when using complex analytical and simulation models. Software has been developed in Python 3.12 and C/C++ 20 to study the proposed changes to the method. With parallel calculations, it is possible to separate the matrix modification of the method into SIMD and multiple-instruction multiple-data (MIMD) components and perform calculations on the appropriate equipment. According to the results of this research, when solving the problem of optimizing benchmark functions of various dimensions, it was possible to accelerate the method by more than 12 times on matrix SIMD central processing unit (CPU) accelerators. When calculating on the graphics processing unit (GPU), the acceleration was about six times due to the difficulties of implementing a pseudo-random number stream. The developed modifications were used to determine the optimal values of the SARIMA parameters when forecasting the volume of transportation by airlines of the Russian Federation. Mathematical dependencies of the acceleration factors on the algorithm parameters and the number of components were also determined, which allows us to estimate the possibilities of accelerating the algorithm by using a reconfigurable heterogeneous computer. Full article

Authenticated encryption with associated data (AEAD) schemes based on stream ciphers, such as ASCON and MORUS, typically use nonlinear feedback shift registers (NFSRs) and linear feedback shift registers (LFSRs) to generate variable-length key streams. While these methods ensure message confidentiality and authenticity, they present challenges in security analysis, especially when automated evaluation is involved. In this paper, we present MLOL, a novel AEAD algorithm based on the LOL framework. MLOL combines authenticated encryption with optimizations to the LFSR structure to enhance both security and efficiency. The cost evaluation demonstrates that on specialized CPU platforms without SIMD instruction set support, MLOL achieves better performance in authenticated encryption speed compared to LOL-MINI with GHASH. Our security analysis confirms that MLOL provides 256-bit security against current cryptanalytic techniques. Experimental results demonstrate that MLOL not only inherits the excellent performance of LOL but also reduces the time complexity of the authenticated encryption process, providing more reliable security guarantees. It significantly simplifies security evaluation, making it suitable for automated analysis tools, and offers a feasible new approach for AEAD algorithm design. Full article

The advancement of Arrival MANager (AMAN) is crucial for addressing the increasing complexity and demand of modern airspace. This study evaluates the operational feasibility and effectiveness of an innovative AMAN designed for en route airspace, the so-called En Route AMAN. The En Route AMAN functions as a controller support system, facilitating the sharing of information between en route air traffic controllers (ATCos), approach controllers (current AMAN), and airport controllers (Departure Managers) in airports with multiple runways. The En Route AMAN aims to support upstream ATCos by sequencing and spacing of incoming streams via speed control and runway assignment, thereby enhancing overall air traffic efficiency. Human-In-The-Loop simulations involving rated ATCos are performed under scenarios that replicate real-world traffic and weather conditions. These simulations focus on upstream airspace to assess the impact of En Route AMAN on delay mitigation and ATCos’ performance. Unlike previous studies that solely relied on theoretical models and fast-time simulation for operational feasibility evaluation, this approach incorporates ATCos’ real-time decision-making, situational awareness, and task management, addressing critical operationalization challenges. The results demonstrated that the En Route AMAN could reduce the average flight duration by up to 25.6 s and decrease the total number of ATCo instructions by up to 20% during peak traffic volume. These findings support that the En Route AMAN is both operationally viable and effective in mitigating arrival delays, highlighting the importance of Human-In-The-Loop for practical validation. Full article

The growing demand for high-performance and energy-efficient processing in edge-oriented Systems-on-Chip is driving the adoption of dedicated integrated circuits that accelerate computationally intensive workloads. To minimize area and performance overhead, low-power, general-purpose CPUs are often tightly coupled with domain-specific coprocessors implementing custom instructions, thereby delivering higher throughput and reduced memory traffic. However, commonly used in-order CPUs are not optimized for instruction-level parallelism, leading to stalls in the instruction stream while waiting for long-latency coprocessor operations and under-utilization of the coprocessor while executing other instructions. This work investigates the benefits of replacing simple in-order cores with a more complex out-of-order architecture to dynamically schedule instructions for the main core and coprocessor, optimizing resource utilization and reducing execution time. To ensure generality, an in-depth analysis was carried out by offloading instructions to a custom dummy coprocessor capable of emulating iterative and pipelined operations with arbitrary latency. Various workloads simulating real-world applications were executed on two variants of an open-source microcontroller equipped with a recent out-of-order core and the state-of-the-art CV32E40X in-order core, respectively. Results from Register Transfer Level simulations show that the former configuration executes up to 60% more instructions per cycle, with a modest 12% system area overhead on a 65 nm CMOS technology node. Full article