The current intrusion detection methods suffer from deficiencies in terms of cross-domain adaptability, privacy preservation, and limited effectiveness in detecting minority-class attacks. To address these issues, a novel intrusion detection model framework, TrMulS, is proposed that integrates federated learning, generative adversarial networks with multispace feature enhancement ability, and transformers with multi-source transfer ability. First, at each institution (source domain), local spatial features are extracted through a CNN, multiple subsets are constructed (to solve the feature singularity problem), and the multihead self-attention mechanism of the transformer is utilized to capture the correlation of features. Second, the synthetic samples of the target domain are generated on the basis of the improved Exchange-GAN, and the cross-domain transfer module is designed by combining the Maximum Mean Discrepancy (MMD) to minimize the feature distribution difference between the source domain and the target domain. Finally, the federated transfer learning strategy is adopted. The model parameters of each local institution are encrypted and uploaded to the target server and then aggregated to generate the global model. These steps iterate until convergence, yielding the globally optimal model. Experiments on the ISCX2012, KDD99 and NSL-KDD intrusion detection standard datasets show that the detection accuracy of this method is significantly improved in cross-domain scenarios. This paper presents a novel paradigm for cross-domain security intelligence analysis that considers efficiency, privacy and balance.

As the manufacturing sector increasingly emphasizes production efficiency and product quality, predicting cutting tool remaining useful life (RUL) and conducting failure analysis become essential for ensuring continuity and reducing costs. However, the uncertainties in the tool wear process complicate accurate predictions and decision-making. This paper introduces a predictive maintenance decision-making approach using an ensemble model of convolutional neural network and bidirectional long short-term memory quantile regression (CNN-BiLSTMQR), enhanced by an attention mechanism and kernel density estimation (KDE). Firstly, signals related to tool wear are collected from various sensors, and a CNN captures spatial features and local patterns in the data. These features are then input into a BiLSTM network, which integrates a temporal attention mechanism and QR to predict RULs at multiple quantiles. Subsequently, KDE generates the probability density distribution of the tool RULs. Finally, the costs and time associated with maintenance decisions are assessed to derive cost functions for tool replacement and ordering. By minimizing the two functions, the optimal timing for tool replacement and spare parts procurement is identified. Validation with a publicly available tool wear prediction dataset demonstrates that the proposed method effectively reduces maintenance costs while ensuring the safe and reliable operation of tools.

ABSTRACT Sparse Matrix‐Vector Multiplication (SpMV) is a fundamental operation in scientific computing, machine learning, and data analysis. The performance of SpMV on GPUs is crucial for accelerating various applications. However, the efficiency of SpMV on GPUs is significantly affected by irregular memory access patterns, high memory bandwidth requirements, and insufficient exploitation of parallelism. In this paper, we propose a Recursive Hybrid Compression (RHC) method to address these challenges. RHC begins by splitting the initial matrix into two portions: an Ellpack (ELL) portion and a Coordinate (COO) portion. This partitioning is followed by further recursive division of the COO portion into additional ELL and COO portions, continuing this process until predefined termination criteria, based on a percentage threshold of the number of nonzero elements, are met. Additionally, we introduce a dynamic partitioning method to determine the optimal threshold for partitioning the matrix into ELL and COO portions based on the distribution of nonzero elements and the memory footprint. We develop the RHC algorithm to fully exploit the advantages of the ELL kernel on GPUs and achieve high thread‐level parallelism. We evaluated our proposed method on two different NVIDIA GPUs: the GeForce RTX 2080 Ti and the A100, using a set of sparse matrices from the SuiteSparse Matrix Collection. We compare RHC with NVIDIA's cuSPARSE library and three state‐of‐the‐art methods: SELLP, MergeBase, and BalanceCSR. RHC achieves average speedups of 2.13, 1.13, 1.87, and 1.27 over cuSPARSE, SELLP, MergeBase, and BalanceCSR, respectively.

In wireless sensor networks(WSNs), data aggregation effectively reduces network traffic, thereby reducing energy consumption and improving network life cycle. Nevertheless, in the process of data aggregation scheduling, if there are fault nodes, the data quality collected by the whole network will decline, and the network performance will decrease, even posing a threat to network security or causing network paralysis. Thus, an efficient and high-performance WSNs restoration algorithm is proposed based on fat tree(FT), which is referred to as the EPRA-FT algorithm. And our goal is to improve the universality, efficiency, and performance retention of the algorithm. Previously, we have conducted a range of the relevant researches on performance improvement of WSNs aggregation scheduling by adopting FT structure, and some successful results have been obtained. On the basis of these results, for the EPRA-FT algorithm, first and foremost, the relationship among nodes is comprehensively recorded in FT construction process. Then, fault nodes are shielded by deleting the known nodes in aggregation tree. Finally, the local reconfiguration of aggregation tree is completed quickly and efficiently. Meanwhile, the aggregation scheduling performance of the original network is maintained to the maximum extent. The feasibility and superiority of our proposed EPRA-FT algorithm are proved by simulation experiments.

With the increasing expansion and complexity of cyberspace data and traffic, network security threats have also increased sharply. As one of the important means to ensure the security of information systems, intrusion detection is facing unprecedented challenges. In this paper, we propose a federated deep transfer learning algorithm, transfer-enhanced deep and transfer domain adaptation (TEDTDA), for intrusion detection. TEDTDA uses federated learning to train local models using intrusion detection data from multiple organizations to protect data privacy. It improves the efficiency of model training by integrating transfer learning theory and knowledge transfers. Moreover, it eliminates unreliable and low-quality local models through model selection in the training process to improve the detection effect. The algorithm is tested on three intrusion detection datasets: ISCX2012, NSL-KDD, and CICIDS2017. Compared with the benchmark algorithm, the proposed TEDTDA algorithm significantly improves the detection accuracy, training efficiency, and other key performance indicators.

As wireless communication technology evolves, precise and swift channel fading prediction is crucial for enhancing communication quality. Traditional methods, however, struggle with computational complexity and accuracy, failing to meet modern systems' real-time and reliability demands. Recently, deep learning has emerged as a promising solution to address the challenge of channel fading prediction. This paper introduces a novel approach, the Blockchain-based Deep Belief Network (BDBN), which leverages blockchain's decentralization, transparency, and security to ensure data integrity and trust during transmission. The BDBN aims to achieve rapid short-term channel fading prediction in wireless communications, thereby optimizing system performance. Meanwhile, the advantages of DBN in dealing with complex nonlinear problems are combined in order to model and predicti wireless channel fading. By training a large amount of historical channel data, the BDBN is able to learn the potential patterns of channel changes, thus realizing highly accurate short-term prediction. Experimental results demonstrate that the BDBN method outperforms traditional prediction methods, enhancing both accuracy and speed. In particular, BDBN exhibits stronger robustness and adaptability when dealing with bursty channel fading. In addition, the introduction of blockchain technology also effectively improves the level of data security and privacy protection, providing a strong guarantee for the secure transmission of wireless communication systems.

Infrared and visible image fusion (IVIF) aims to generate fused images with prominent targets and rich scene information. However, in low-light conditions, visible images lose accurate texture and color, reducing their ability to provide detailed scene information for fusion. Existing IVIF methods often overlook illumination degradation and cause color distortion when incorporating infrared information. To address these problems, we propose a novel visually enhanced IVIF method tailored for low-light environments. Our method combines low-light image enhancement (LLIE) and IVIF into a single module. First, we adaptively enhance the low-light visible image, ensuring rich texture and color for fusion. Additionally, we introduce a three-channel fusion coefficient map to transform infrared information into visible image, preventing color distortion and highlighting key targets while maintaining details of the fused image. Since infrared and visible images are from different modalities, we map them into the same high-dimensional feature space. We then propose the feature difference to integrate complementary information, producing a fused image with complete content and no redundancy. Notably, our method is zero-shot, requiring only a pair of test infrared and visible images for training. This better meets the complexity of IVIF in various low-light scenes. Extensive experiments show that in low-light conditions, our method surpasses other state-of-the-art (SOTA) methods by providing more natural colors, richer textures, and better alignment with human visual perception.

Sparse matrix–vector multiplication (SpMV) is a computationally intensive kernel ubiquitously employed in various linear algebra computations. In numerous high-performance applications, SpMV computations often become a performance bottleneck, making their optimization crucial. However, the irregular structure of sparse matrices can lead to redundant calculations, resulting in unbalanced loads and inefficient bandwidth utilization during SpMV operations on GPUs. In this paper, we introduce a novel approach called block strategy and adaptive storage (BSAS) to address these challenges. Our methodology involves partitioning the original matrix into uniform-distribution blocks based on the block strategy, ensuring maximum similarity in the count of non-zero elements per block. This partitioning balances the load, reduces redundant calculations, and minimizes conversion overhead. Additionally, we dynamically assign storage formats to each block using an adaptive storage format selection method that considers the distribution of non-zero elements within the blocks. Furthermore, we develop optimized kernel implementations for each block, significantly enhancing thread-level parallelism and memory utilization. We evaluated our proposed method against NVIDIA’s cuSPARSE library and three advanced SpMV algorithms: SELLP, Merge-Based, and BalanceCSR, using matrices from the SuiteSparse dataset on NVIDIA RTX 2080 Ti and A100 GPUs. The results indicated that our method outperformed the compared algorithms, achieving average speedups of 1.84× , 1.30× , 1.94× , and 1.42× over cuSPARSE, SELLP, Merge-Based, and BalanceCSR, respectively.

Oriented remote sensing object detection (ORSOD) has gained increasing significance in both military and civilian applications due to the necessity of accurately identifying objects with varying shapes and orientations in remote sensing data. Traditional ORSOD methods often employ fixed label assignment strategies to differentiate between positive and negative samples. However, most of them frequently overlook the impact of object shape on sample quality, leading to an imbalanced distribution of positive samples and exacerbating the inconsistency between classification and regression tasks, thereby limiting detection performance. To address these challenges, we propose a novel Shape-Dependent Assignment (SDA) method that dynamically differentiates positive and negative samples based on object shape. It introduces a new metric for evaluating sample box quality by considering angular differences relative to ground truth (GT) boxes and adjusts the sample scoring threshold according to the aspect ratio of each GT box. In addition, we present a DIoU-Adaptive Weighting (DAW) module that enhances the interaction between classification and regression tasks by leveraging the Distance-IoU metric. This approach not only balances the quantity of samples but also improves their quality, enabling more effective training schemes for samples of varying qualities. We validate our proposed methods through extensive experiments on three challenging ORSOD datasets: DOTA-1.0, HRSC2016, and UCAS-AOD. The results demonstrate that our approach achieves significant improvements, especially for objects with large aspect ratios.

Binary Neural Networks (BNNs) represent an extreme form of neural network quantization, achieving compact storage, low energy consumption, and fast computation by restricting network weights to -1, +1 . In BNNs, parameter updates involve flipping weights between -1 and +1 , meaning that only the sign of the weights matters, not their magnitude. This makes the optimization process for BNNs fundamentally different from that of full-precision neural networks, leading to the development of optimizers tailored specifically for BNNs. The current mainstream approach involves using a fixed threshold to suppress frequent flipping of network parameters, thereby improving the stability of updates and enhancing convergence speed and performance. However, since this fixed threshold is a pre-set hyperparameter, it introduces additional complexity to the training process and reduces flexibility. In this paper, we experimentally explore the impact of threshold variation on training outcomes, discuss the relationship between the learning rate and the threshold, and propose the Binary Optimizer with Adaptive Thresholds (BOAT). In BOAT, we eliminate the learning rate—a common hyperparameter in most optimization algorithms—and instead, use the accumulation of gradient squares of the weights to implement an adaptive, incrementally adjusting threshold. This reduces the number of hyperparameters, simplifies the training process, and results in better convergence and generalization. Additionally, we provide a thorough ablation study of the hyperparameter space and analyze the effects of weight decay and threshold growth rate on performance. We validate BOAT through image classification experiments on the CIFAR-10, CIFAR-100, and TinyImageNet datasets using the BinaryNet and ResNet-18 architectures. Our experiments demonstrate that BOAT consistently outperforms other binary optimizers. The code is available at: https://github.com/gushan/BOAT .

Existing methods for infrared and visible image fusion (IVIF) effectively highlight thermal targets but often fail to accurately and fully represent scene information, particularly in low-light environments. This limitation stems from two primary reasons: the modality difference between infrared and visible images, which can distort the fused image when introducing single-channel infrared information, and the ignorance of illumination degradation in the visible image, which can make the fused image appear dark and lacking in scene details. To address these challenges, we propose an efficient feature difference-based IVIF method tailored for low-light conditions. Our method jointly processes low-light image enhancement (LLIE) and IVIF. First, the low-light visible image is adaptively enhanced to produce a bright scene with rich textures and colors. Additionally, we map infrared and visible images into a same high-dimensional feature space and present a feature difference mechanism to ensure the fused image contains all visible image information while retaining only infrared image unique information. This avoids the risk of distortion from introducing redundant infrared information. We also propose a three-channel fusion coefficient map and strengthen the correlation between infrared and visible modalities, converting infrared unique information into a three-channel format. This resolves distortion issues from modality differences during fusion, ensuring accurate scene details. To achieve low computational cost and high efficiency during testing, we employ a multi-stage training strategy, enabling a single lightweight module to produce stable fusion results. Extensive experiments demonstrate that our method surpasses state-of-the-art (SOTA) methods in terms of fusion quality and efficiency.

In large-scale data analysis, efficient Top-K query processing is critical for numerous applications in science, industry, and society. Traditional approaches often involve substantial data transfer and computational overhead, making it difficult to meet the scalability and efficiency demands of modern datasets. This paper proposes a GPU-accelerated Top-K query processing method that integrates data compression and pre-filtering techniques to address these challenges. By partitioning and compressing data on the host side, it alleviates common PCIe bottlenecks in heterogeneous computing environments. A metadata-driven pre-filtering technique further reduces the data volume processed on the GPU, significantly improving query performance, particularly when handling anti-correlated datasets. Experimental results demonstrate that this method markedly reduces data transfer and processing time, confirming its effectiveness in enhancing the efficiency and scalability of Top-K query processing compared to existing methods.

Top-K and selection operations are critical in data processing and analysis, and their efficient implementation on GPUs is increasingly important due to the growing demands of data analysis. Existing methods, primarily relying on the bucket partition execution model, encounter challenges such as uneven bucket distribution and latency in merging processes. To address these issues, we introduce a novel Split-Bucket Partition (SBP) execution model that specifically addresses these challenges. Additionally, we propose task and control flow optimizations targeted at top-K and selection algorithms, which further contribute to performance improvements. Our optimized algorithms significantly outperform existing approaches, delivering performance gains of up to 2.3 times and 1.6 times for different bucket partitioning rules. Our algorithms show robust performance improvements in non-uniform data scenarios, with gains ranging from 1.9 times to 15.5 times. However, it should be noted that the SBP model has limitations related to shared memory and register utilization, potentially impacting performance. Tests on TU102 and A100 GPU architectures validate the effectiveness of our approach, achieving a maximum speedup of 2.9 times. The study suggests that while the SBP model is effective for top-K and selection algorithms, it also holds promise for other computational tasks, setting the stage for future research.

With advancements in computing technology and the rapid progress of data science, machine learning has been widely applied in various fields, showing great potential, especially in digital healthcare. In recent years, conversational diagnostic systems have been used to predict diseases through symptom checking. Early systems predicted the likelihood of a single disease by minimizing the number of questions asked. However, doctors typically perform differential diagnoses in real medical practice, considering multiple possible diseases to address diagnostic uncertainty. This requires systems to ask more critical questions to improve diagnostic accuracy. Nevertheless, such systems in acute medical situations need to process information quickly and accurately, but the complexity of differential diagnosis increases the system’s computational cost. To improve the efficiency and accuracy of telemedicine diagnostic systems, this study developed an optimized algorithm for the Top-K algorithm. This algorithm dynamically adjusts the number of the most likely diseases and symptoms by real-time monitoring of case progress, optimizing the diagnostic process, enhancing accuracy (99.81%), and increasing the exclusion rate of severe pathologies. Additionally, the Top-K algorithm optimizes the diagnostic model through a policy network loss function, effectively reducing the number of symptoms and diseases processed and improving the system’s response speed by 1.3–1.9 times compared to the state-of-the-art differential diagnosis systems.

Human activity recognition(HAR) is an important research focus in ubiquitous computing. It has been widely applied in various domains, such as smart homes, healthcare assistance, and sports training. Accurate human activity recognition directly impacts the performance of downstream tasks. Existing methods for human activity recognition primarily rely on extracting temporal or spatial features from the data. These features are used for the task of human activity recognition. The existing methods mainly use CNN or RNN models to model the Euclidean correlations among spatially adjacent sensors or channels. However, non-Euclidean pairwise correlations among all sensors or channels are even critical for accurate classification, which has been ignored by the existing methods. In this paper, we incorporate fine-grained spatial structural data into the model to overcome these limitations. A novel deep learning model for human activity recognition is proposed, which is called the fine-grained data-oriented Spatial-Temporal Graph Transformer network (STGT). The introduction of the STGT model can eliminate the limitations of existing spatial feature extraction methods by leveraging a novel data organization approach proposed in this study. This model enhances the spatial features within the time feature extraction module for effective spatio-temporal feature extraction. We conducted experiments on four large-scale real-world HAR datasets to evaluate its performance. The experimental results demonstrate the superiority of our method over state-of-the-art approaches.

Intrusion detection technology is an active network defense technology that has quickly become an important means of ensuring network security since it was proposed. With the development of machine learning, its application in intrusion detection has effectively improved the detection effect. However, due to the increasing volume and dimension of the current network data, the model training speed of machine learning-based intrusion detection technology is slow, and the update cost is high. In addition, different distributions of intrusion detection data affect the detection effect. In this paper, the extreme learning machine (ELM) model is extended. An intrusion detection method, TCA-ELM, based on transfer component analysis (TCA) and an extreme learning machine is proposed. TCA-ELM utilizes TCA to extract data features and project data to a new space by greatly reducing the distance and dimension between the data distributions and then trains the ELM in the new feature space to improve the training efficiency and detection rate. Experimental tests are performed on the famous NSL-KDD dataset, and the experimental results demonstrate the effectiveness of the proposed TCA-ELM method. Compared with the other four benchmark algorithms, the training time is significantly improved, and the detection accuracy is also significantly improved.

In the current data sharing environment of smart manufacturing systems, there is a lack of a transparent, open, and fair interaction environment, and there exists a serious imbalance in data sharing and transmission, which leads to data trust issues. To address this challenge, we propose a fine-grained access control method for IoT data sharing based on blockchain technology (the Bayesian information criterion [BIC] scheme for IoT data sharing). This approach includes ex-ante supervision and authentication of data sources, fine-grained access control during data usage, and ex-post tracking of malicious users. Firstly, a system model is built, and symmetric encryption is used to encrypt the data in order to improve computational efficiency. Subsequently, scheme design and algorithm design are carried out to encrypt the symmetric key using ciphertext-policy attribute-based encryption to achieve one-to-many data sharing, which greatly reduces the number of repeated encryptions and computational overheads. The data owner uploads the encrypted data to the cloud server, and the data user must hold the token generated by the blockchain to interact with the cloud and obtain the corresponding information. The results show that the BIC scheme not only ensures the legitimacy and security of data access but also locks up and traces malicious users who may have compromised their private keys, preventing illegal data sharing and key misuse; it offers significant advantages in terms of storage efficiency and time consumption, and also facilitates a flexible and secure data sharing process.

Unmanned Aerial Vehicles (UAVs) have been widely used in emergency rescue because of their small size and low cost, which can effectively reduce casualties and improve rescue efficiency. However, for UAV collision avoidance path planning in an unknown dynamic environment is a relatively difficult optimization problem. On the one hand, it is difficult for traditional algorithms to adapt to the huge state space generated by three-dimensional environment, which makes convergence difficult. On the other hand, the dynamic changing environment will affect the flight safety of UAVs. In this paper, an Improved Memory Pool Dueling Deep Q Network algorithm (IM-Dueling DQN) is proposed to optimize obstacle avoidance and path planning. Combined with the state and action space, a comprehensive reward function is designed to speed up the convergence of the algorithm and improve the generalization ability of the UAV to the environment. Experimental simulations show that the algorithm has good robustness in an unknown dynamic environment, and can avoid all obstacles and plan a safe navigation route.

Sparse matrix-vector multiplication (SpMV) plays a critical role in a wide range of linear algebra computations, particularly in scientific and engineering disciplines. However, the irregular memory access patterns, extensive memory usage, high bandwidth requirements, and underutilization of parallelism hinder the computational efficiency of SpMV on GPUs. In this paper, we propose a novel approach called block-wise dynamic mixed-precision (BDMP) to address these challenges. Our methodology involves partitioning the original matrix into uniformly sized blocks, with each block’s size determined by considering architectural characteristics and accuracy requirements. Additionally, we dynamically assign precision to each block using a precision selection method that takes into account the value distribution of the original sparse matrix. We develop two distinct SpMV computation algorithms for BDMP: BDMP-PBP (Precision-based partitioning) and BDMP-TCKI (Tailored compression and kernel implementation). BDMP-PBP partitions the matrix into two independent matrices for separate computations based on block precision, offering flexibility for integration with other optimization techniques. Meanwhile, BDMP-TCKI focuses on achieving significant thread-level parallelism and memory utilization by tailoring an appropriate compressed storage format and kernel implementation for each block. We compare BDMP with NVIDIA’s cuSPARSE library and three state-of-the-art SpMV methods, including SELLP, MergeBase, and BalanceCSR, using matrices from the University of Florida’s SuiteSparse dataset collection. BDMP-PBP and BDMP-TCKI show average speedups up to 2.64 × and 2.91 × on Turing RTX 2080Ti, and up to 2.99 × and 3.22 × on Ampere A100. The results demonstrate that BDMP enables the optimization of computation speed without compromising the precision necessary for reliable results.

1 Introduction The Internet of Things(IoT)has facilitated the development of numerous fields in our lives.However,some equipment in IoT environment lacks sufficient storage and data processing capabilities[1].A feasible strategy is to leverage the powerful computing capabilities of cloud servers to process the data within the IoT devices.