Search Results (148)

The proliferation of biometric technology across various domains including user identification, financial services, healthcare, security, law enforcement, and border control introduces convenience in user identity verification while necessitating robust protection mechanisms for sensitive biometric data. While chaos-based encryption systems offer promising solutions, many existing chaos-based encryption schemes exhibit inherent shortcomings including deterministic randomness and constrained key spaces, often failing to balance security robustness with computational efficiency. To address this, we propose a novel dual-layer cryptographic framework leveraging a four-dimensional (4D) Qi hyperchaotic system for protecting biometric templates and facilitating secure feature matching operations. The framework implements a two-tier encryption mechanism where each layer independently utilizes a Qi hyperchaotic system to generate unique encryption parameters, ensuring template-specific encryption patterns that enhance resistance against chosen-plaintext attacks. The framework performs dimensional normalization of input biometric templates, followed by image pixel shuffling to permutate pixel positions before applying dual-key encryption using the Qi hyperchaotic system and XOR diffusion operations. Templates remain encrypted in storage, with decryption occurring only during authentication processes, ensuring continuous security while enabling biometric verification. The proposed system’s framework demonstrates exceptional randomness properties, validated through comprehensive NIST Statistical Test Suite analysis, achieving statistical significance across all 15 tests with p-values consistently above 0.01 threshold. Comprehensive security analysis reveals outstanding metrics: entropy values exceeding 7.99 bits, a key space of ${10}^{320}$, negligible correlation coefficients (<${10}^{-2}$), and robust differential attack resistance with an NPCR of 99.60% and a UACI of 33.45%. Empirical evaluation, on standard CASIA Face and Iris databases, demonstrates practical computational efficiency, achieving average encryption times of 0.50913s per user template for 256 × 256 images. Comparative analysis against other state-of-the-art encryption schemes verifies the effectiveness and reliability of the proposed scheme and demonstrates our framework’s superior performance in both security metrics and computational efficiency. Our findings contribute to the advancement of biometric template protection methodologies, offering a balanced performance between security robustness and operational efficiency required in real-world deployment scenarios. Full article

Accurate modeling of ionospheric variability is critical for space weather forecasting and GNSS applications. While machine learning approaches have shown promise, progress is hindered by the absence of standardized benchmarking practices and narrow test periods. In this paper, we take the first step toward fostering rigorous and reproducible evaluation of AI models for ionospheric forecasting by introducing IonoBench: a benchmarking framework that employs a stratified data split, balancing solar intensity across subsets while preserving 16 high-impact geomagnetic storms (Dst ≤ $-100$ nT) for targeted stress testing. Using this framework, we benchmark a field-specific model (DCNN) against state-of-the-art spatiotemporal architectures (SwinLSTM and SimVPv2) using the climatological IRI 2020 model as a baseline reference. DCNN, though effective under quiet conditions, exhibits significant degradation during elevated solar and storm activity. SimVPv2 consistently provides the best performance, with superior evaluation metrics and stable error distributions. Compared to the C1PG baseline (the CODE 1-day forecast product), SimVPv2 achieves a notable RMSE reduction up to $32.1\%$ across various subsets under diverse solar conditions. The reported results highlight the value of cross-domain architectural transfer and comprehensive evaluation frameworks in ionospheric modeling. With IonoBench, we aim to provide an open-source foundation for reproducible comparisons, supporting more meticulous model evaluation and helping to bridge the gap between ionospheric research and modern spatiotemporal deep learning. Full article

Intrusion detection systems are crucial for cybersecurity applications. Network traffic data originate from diverse terminal sources, exhibiting multi-view feature spaces, while the collection of unknown intrusion data is costly. Current one-class classification (OCC) approaches are mainly designed for single-view data. Multi-view OCC approaches usually require collecting multi-view traffic data from all sources and have difficulty detecting intrusion independently in each view. Furthermore, they commonly ignore the potential subcategories in normal traffic data. To address these limitations, this paper utilizes the Broad Learning System (BLS) technique and proposes an intrusion detection framework based on a multi-view cluster structure guided one-class BLS-autoencoder (IDF-MOCBLSAE). Specifically, a multi-view co-association matrix optimization objective function with doubly-stochastic constraints is first designed to capture the cross-view cluster structure. Then, a multi-view cluster structure guided one-class BLS-autoencoder (MOCBLSAEs) is proposed, which learns the discriminative patterns of normal traffic data by preserving the cross-view clustering structure while minimizing the intra-view sample reconstruction errors, thereby enabling the identification of unknown intrusion data. Finally, an intrusion detection framework is constructed based on multiple MOCBLSAEs to achieve both individual and ensemble intrusion detection. Through experimentation, IDF-MOCBLSAE is validated on real-world network traffic datasets for multi-view one-class classification tasks, demonstrating its superiority over state-of-the-art one-class approaches. Full article

The transformer-based deep learning approach represents the current state-of-the-art in machine translation (MT) research. Large-scale pretrained transformer models produce state-of-the-art performance across a wide range of MT tasks for many languages. However, such deep neural network (NN) models are often data-, compute-, space-, power-, and energy-hungry, typically requiring powerful GPUs or large-scale clusters to train and deploy. As a result, they are often regarded as “non-green” and “unsustainable” technologies. Distilling knowledge from large deep NN models (teachers) to smaller NN models (students) is a widely adopted sustainable development approach in MT as well as in broader areas of natural language processing (NLP), including speech, and image processing. However, distilling large pretrained models presents several challenges. First, increased training time and cost that scales with the volume of data used for training a student model. This could pose a challenge for translation service providers (TSPs), as they may have limited budgets for training. Moreover, CO₂ emissions generated during model training are typically proportional to the amount of data used, contributing to environmental harm. Second, when querying teacher models, including encoder–decoder models such as NLLB, the translations they produce for low-resource languages may be noisy or of low quality. This can undermine sequence-level knowledge distillation (SKD), as student models may inherit and reinforce errors from inaccurate labels. In this study, the teacher model’s confidence estimation is employed to filter those instances from the distilled training data for which the teacher exhibits low confidence. We tested our methods on a low-resource Urdu-to-English translation task operating within a constrained training budget in an industrial translation setting. Our findings show that confidence estimation-based filtering can significantly reduce the cost and CO₂ emissions associated with training a student model without drop in translation quality, making it a practical and environmentally sustainable solution for the TSPs. Full article

This study explores the differences in audience behavior between virtual museums and physical museums. The replica-based virtual museum (RVM) was developed to replicate the exhibit layout of physical museums and support multi-user online visits. The study introduces the RVM-Interaction (RVM-I), which incorporates interactive features to enhance user engagement. In the experiment, 24 participants experienced a physical museum (PM), RVM, RVM-I, and a traditional PC-based virtual museum, with their impressions and behavioral patterns recorded. The results indicate no significant differences between RVM and PM in terms of satisfaction, immersion, aesthetic experience, and social interaction. RVM-I significantly enhanced the participants’ experience through its interactive capabilities. Path analysis shows that both RVM and RVM-I improved audience efficiency, with RVM-I transforming the circumferential, space-based art appreciation found in PM and RVM into a stationary, space-based form, making RVM-I more engaging than RVM. These findings offer valuable insights for the design and development of virtual museum experiences that maintain spatial fidelity to physical exhibitions while enhancing user engagement through interactivity. Full article

Advancements in CMOS technology have significantly reduced both transistor dimensions and inter-device spacing, leading to a lower critical charge at sensitive nodes. As a result, SRAM cells used in space applications have become increasingly vulnerable to single-event upset (SEU) caused by the harsh radiation environment. To ensure reliable operation under such conditions, radiation-hardened SRAM designs are essential. In this paper, we propose a low-power read-decoupled radiation-hardened 16T (LDRH16T) SRAM cell to mitigate the effects of SEU. The proposed cell is evaluated against several state-of-the-art soft-error-tolerant SRAM designs, including QUCCE12T, WE-QUATRO, RHBD10T, SIS10T, EDP12T, SEA14T, and SAW16T. Simulations are conducted using a 90 nm CMOS process at a supply voltage of 1 V and a temperature of 27 °C. Simulation results show that LDRH16T successfully recovers its original state after injection at all sensitive nodes. Furthermore, since its storage nodes are decoupled from the bit lines during read operations, the proposed cell achieves the highest read stability among the compared designs. It also exhibits superior write ability, shorter write delay, and significantly lower hold power consumption. In addition, LDRH16T demonstrates excellent overall performance across key evaluation metrics and proves its capability for reliable operation in space environments. Full article

In recent years, Large Language Models (LLMs) have witnessed rapid advancements, revolutionizing various domains. Within the realm of software development, code generation technology powered by LLMs has emerged as a prominent research focus. Despite its potential, the application of this technology in the aerospace sector remains in its nascent, exploratory phase. This paper delves into the intricacies of LLM-based code generation methods and explores their potential applications in aerospace contexts. It introduces RepoSpace, the pioneering warehouse-level benchmark test for code generation of spaceborne equipment. Comprising 825 samples from five actual projects, this benchmark offers a more precise evaluation of LLMs’ capabilities in aerospace scenarios. Through extensive evaluations of seven state-of-the-art LLMs on RepoSpace, the study reveals that domain-specific differences significantly impact the code generation performance of LLMs. Existing LLMs exhibit subpar performance in specialized warehouse-level code generation tasks for aerospace, with their performance markedly lower than that of domain tasks. The research further demonstrates that Retrieval Augmented Generation (RAG) technology can effectively enhance LLMs’ code generation capabilities. Additionally, the use of appropriate prompt templates can guide the models to achieve superior results. Moreover, high-quality documentation strings are found to be crucial in improving LLMs’ performance in warehouse-level code generation tasks. This study provides a vital reference for leveraging LLMs for code generation in the aerospace field, thereby fostering technological innovation and progress in this critical domain. Full article

The contemporary challenge in remote sensing lies in the precise retrieval of increasingly abundant and high-resolution remotely sensed images (RS image) stored in expansive data warehouses. The heightened spatial and spectral resolutions, coupled with accelerated image acquisition rates, necessitate advanced tools for effective data management, retrieval, and exploitation. The classification of large-sized images at the pixel level generates substantial data, escalating the workload and search space for similarity measurement. Semantic-based image retrieval remains an open problem due to limitations in current artificial intelligence techniques. Furthermore, on-board storage constraints compel the application of numerous compression algorithms to reduce storage space, intensifying the difficulty of retrieving substantial, sensitive, and target-specific data. This research proposes an innovative hybrid approach to enhance the retrieval of remotely sensed images. The approach leverages multilevel classification and multiscale feature extraction strategies to enhance performance. The retrieval system comprises two primary phases: database building and retrieval. Initially, the proposed Multiscale Multiangle Mean-shift with Breaking Ties (MSMA-MSBT) algorithm selects informative unlabeled samples for hyperspectral and synthetic aperture radar images through an active learning strategy. Addressing the scaling and rotation variations in image capture, a flexible and dynamic algorithm, modified Deep Image Registration using Dynamic Inlier (IRDI), is introduced for image registration. Given the complexity of remote sensing images, feature extraction occurs at two levels. Low-level features are extracted using the modified Multiscale Multiangle Completed Local Binary Pattern (MSMA-CLBP) algorithm to capture local contexture features, while high-level features are obtained through a hybrid CNN structure combining pretrained networks (Alexnet, Caffenet, VGG-S, VGG-M, VGG-F, VGG-VDD-16, VGG-VDD-19) and a fully connected dense network. Fusion of low- and high-level features facilitates final class distinction, with soft thresholding mitigating misclassification issues. A region-based similarity measurement enhances matching percentages. Results, evaluated on high-resolution remote sensing datasets, demonstrate the effectiveness of the proposed method, outperforming traditional algorithms with an average accuracy of 86.66%. The hybrid retrieval system exhibits substantial improvements in classification accuracy, similarity measurement, and computational efficiency compared to state-of-the-art scene classification and retrieval methods. Full article

Accurate cloud detection is a critical preprocessing step in remote sensing applications, as cloud and cloud shadow contamination can significantly degrade the quality of optical satellite imagery. In this paper, we propose CDWMamba, a novel dual-domain neural network that integrates the Mamba-based state space model with discrete wavelet transform (DWT) for effective cloud detection. CDWMamba adopts a four-direction Mamba module to capture long-range dependencies, while the wavelet decomposition enables multi-scale global context modeling in the frequency domain. To further enhance fine-grained spatial features, we incorporate a multi-scale depth-wise separable convolution (MDC) module for spatial detail refinement. Additionally, a spectral–spatial bottleneck (SSN) with channel-wise attention is introduced to promote inter-band information interaction across multi-spectral inputs. We evaluate our method on two benchmark datasets, L8 Biome and S2_CMC, covering diverse land cover types and environmental conditions. Experimental results demonstrate that CDWMamba achieves state-of-the-art performance across multiple metrics, significantly outperforming deep-learning-based baselines in terms of overall accuracy, mIoU, precision, and recall. Moreover, the model exhibits satisfactory performance under challenging conditions such as snow/ice and shrubland surfaces. These results verify the effectiveness of combining a state space model, frequency-domain representation, and spectral–spatial attention for cloud detection in multi-spectral remote sensing imagery. Full article

Art exhibition spaces increasingly emphasize visitor experience, yet the relationships among spatial structure, visitor behavior, and emotional response remain unclear. Traditional space syntax analyses typically focus on physical spatial structures, insufficiently capturing visitors’ emotional and cognitive experiences. To address these gaps, this study presents an integrative evaluation framework that combines space syntax theory with multimodal physiological measurements to systematically assess spatial design performance in art exhibition environments. Eye-tracking and heart rate variability (HRV) experiments were conducted to investigate how spatial configuration affects visual attention and emotional responses. Visibility graph analysis, spatial integration metrics, and regression modeling were applied using the third-floor temporary exhibition hall of the Pudong Art Museum in Shanghai as a case study. The results revealed that HRV levels (β = −7.92) were significantly predicted via spatial integration, and the relationship between spatial integration and the number of fixations was partially mediated by HRV (indirect effect: β = −0.36; direct effect: β = 8.23). Additionally, zones with higher occlusivity were associated with more complex scanpaths (mean complexity: 0.14), whereas highly integrated regions triggered more fixations (mean = 10.54) and longer total fixation durations (mean = 2946.98 ms). Therefore, spatial syntax, when coupled with physiological indicators, provides a robust and actionable method for evaluating and optimizing exhibition space design. Full article

The latest research indicates that Large Vision-Language Models (VLMs) have a wide range of applications in the field of remote sensing. However, the vast amount of image data in this field presents a challenge in selecting high-quality multimodal data, which are essential for saving computational resources and time. Therefore, we propose an adaptive fine-tuning algorithm for multimodal large models. The core steps of this algorithm involve two stages of data truncation. First, the vast dataset is projected into a semantic vector space, where the MiniBatchKMeans algorithm is used for automated clustering. This classification ensures that the data within each cluster exhibit high semantic similarity. Next, the data within each cluster are processed by calculating the translational difference between the original and perturbed data in the multimodal large model’s vector space. This difference serves as a generalization metric for the data. Based on this metric, we select data with high generalization potential for training. We apply this algorithm to train the InternLM-XComposer2-VL-7B model on two 3090 GPUs, using one-third of the GeoChat multimodal remote sensing dataset. The results demonstrate that our algorithm outperforms state-of-the-art baselines. The model trained on our optimally chosen one-third dataset, as validated through experiments, showed only a 1% reduction in performance across various remote sensing metrics compared to the model trained on the full dataset. This approach significantly preserved general-purpose capabilities while reducing training time by 68.2%. Furthermore, the model achieved scores of 89.86 and 77.19 on the UCMerced and AID evaluation datasets, respectively, surpassing the GeoChat dataset by 5.43 and 5.16 points. It only showed a 0.91-point average decrease on the LRBEN evaluation dataset. Full article

Aerial object detection plays a critical role in numerous fields, utilizing the flexibility of airborne platforms to achieve real-time tasks. Combining visible and infrared sensors can overcome limitations under low-light conditions, enabling full-time tasks. While feature-level fusion methods exhibit comparable performances in visible–infrared multispectral object detection, they suffer from heavy model size, inadequate inference speed and visible light preferences caused by inherent modality imbalance, limiting their applications in airborne platform deployment. To address these challenges, this paper proposes a YOLO-based real-time multispectral fusion framework combining pixel-level fusion with dynamic modality-balanced augmentation called Full-time Multispectral Pixel-wise Fusion Network (FMPFNet). Firstly, we introduce the Multispectral Luminance Weighted Fusion (MLWF) module consisting of attention-based modality reconstruction and feature fusion. By leveraging YUV color space transformation, this module efficiently fuses RGB and IR modalities while minimizing computational overhead. We also propose the Dynamic Modality Dropout and Threshold Masking (DMDTM) strategy, which balances modality attention and improves detection performance in low-light scenarios. Additionally, we refine our model to enhance the detection of small rotated objects, a requirement commonly encountered in aerial detection applications. Experimental results on the DroneVehicle dataset demonstrate that our FMPFNet achieves 76.80% mAP50 and 132 FPS, outperforming state-of-the-art feature-level fusion methods in both accuracy and inference speed. Full article

High-dimensional complex optimization problems are pervasive in engineering and scientific computing, yet conventional algorithms struggle to meet collaborative optimization requirements due to computational complexity. While Chicken Swarm Optimization (CSO) demonstrates an intuitive understanding and straightforward implementation for low-dimensional problems, it suffers from limitations including a low convergence precision, uneven initial solution distribution, and premature convergence. This study proposes an Adaptive Dynamically Enhanced Variant of Chicken Swarm Optimization (ADVCSO) algorithm. First, to address the uneven initial solution distribution in the original algorithm, we design an elite perturbation initialization strategy based on good point sets, combining low-discrepancy sequences with Gaussian perturbations to significantly improve the search space coverage. Second, targeting the exploration–exploitation imbalance caused by fixed role proportions, a dynamic role allocation mechanism is developed, integrating cosine annealing strategies to adaptively regulate flock proportions and update cycles, thereby enhancing exploration efficiency. Finally, to mitigate the premature convergence induced by single update rules, hybrid mutation strategies are introduced through phased mutation operators and elite dimension inheritance mechanisms, effectively reducing premature convergence risks. Experiments demonstrate that the ADVCSO significantly outperforms state-of-the-art algorithms on 27 of 29 CEC2017 benchmark functions, achieving a 2–3 orders of magnitude improvement in convergence precision over basic CSO. In complex composite scenarios, its convergence accuracy approaches that of the championship algorithm JADE within a 10⁻² magnitude difference. For collaborative multi-subproblem optimization, the ADVCSO exhibits a superior performance in both Multiple Traveling Salesman Problems (MTSPs) and Multiple Knapsack Problems (MKPs), reducing the maximum path length in MTSPs by 6.0% to 358.27 units while enhancing the MKP optimal solution success rate by 62.5%. The proposed algorithm demonstrates an exceptional performance in combinatorial optimization and holds a significant engineering application value. Full article

Semantic segmentation of urban meshes plays an increasingly crucial role in the analysis and understanding of 3D environments. Most existing large-scale urban mesh semantic segmentation methods focus on integrating multi-scale local features but struggle to model long-range dependencies across facets effectively. Furthermore, owing to high computational complexity or excessive pre-processing operations, these methods lack the capability for the efficient semantic segmentation of large-scale urban meshes. Inspired by Mamba, we propose MambaMeshSeg-Net, a novel 3D urban mesh semantic segmentation method based on the State Space Model (SSM). The proposed method incorporates a hybrid scanning strategy that adaptively scans 3D urban meshes to extract long-range dependencies across facets, enhancing semantic segmentation performance. Moreover, our model exhibits faster performance in both inference and pre-processing compared to other mainstream models. In comparison with existing state-of-the-art (SOTA) methods, our model demonstrates superior performance on two widely utilized open urban mesh datasets. Full article

Due to the rapid proliferation of social media and online reviews, the accurate identification and classification of controversial texts has emerged as a significant challenge in the field of natural language processing. However, traditional text-classification methodologies frequently encounter critical limitations, such as feature sensitivity and inadequate generalization capabilities. This results in a notably suboptimal performance when confronted with diverse controversial content. To address these substantial limitations, this paper proposes a novel controversial text-detection framework based on stacked ensemble learning to enhance the accuracy and robustness of text classification. Firstly, considering the multidimensional complexity of textual features, we integrate comprehensive feature engineering, i.e., encompassing word frequency, statistical metrics, sentiment analysis, and comment tree structure features, as well as advanced feature selection methodologies, particularly lassonet, i.e., a neural network with feature sparsity, to effectively address dimensionality challenges while enhancing model interpretability and computational efficiency. Secondly, we design a two-tier stacked ensemble architecture, which not only combines the strengths of multiple machine learning algorithms, e.g., gradient-boosted decision tree (GBDT), random forest (RF), and extreme gradient boosting (XGBoost), with deep learning models, e.g., gated recurrent unit (GRU) and long short-term memory (LSTM), but also implements the support vector machine (SVM) for efficient meta-learning. Furthermore, we systematically compare three hyperparameter optimization algorithms, including the sparrow search algorithm (SSA), particle swarm optimization (PSO), and Bayesian optimization (BO). The experimental results demonstrate that the SSA exhibits a superior performance in exploring high-dimensional parameter spaces. Extensive experimentation across diverse topics and domains also confirms that our proposed methodology significantly outperforms the state-of-the-art approaches. Full article