包头职业技术学院

Adversarial attacks pose serious security risks to deep neural networks (DNNs) by inducing erroneous predictions. While gradient-based and GAN-based attack methods have achieved high success rates, they often produce perturbations that are diffusely distributed across images, resulting in poor imperceptibility. This limitation restricts their applicability in scenarios such as privacy protection for face recognition systems. To address this issue, we propose SpAdvGAN, a semantic-aware adversarial example generation framework that integrates sparse attention with generative adversarial networks. SpAdvGAN introduces a sparse semantic attention extractor that explicitly constrains perturbations to a limited set of semantically important regions, preventing unnecessary noise diffusion and preserving visual fidelity. This design enables precise regulation of perturbation granularity while maintaining strong attack effectiveness. Extensive experiments on CIFAR-10 and ImageNet datasets demonstrate that SpAdvGAN substantially improves the imperceptibility of adversarial examples, while achieving comparable or higher attack success rates than state-of-the-art gradient-based attacks and the GAN-based baseline LpAdvGAN.

This paper proposes an adaptive prescribed-time tracking control strategy based on the dynamic surface technique for hydraulic servo systems subject to time-varying parameters, external disturbances, and output constraints. First, a state-constrained transformation function is introduced to convert the strict output constraint condition into an error boundedness problem. Meanwhile, the dynamic surface control (DSC) technique is employed to effectively avoid the “explosion of complexity” inherent in traditional backstepping design. Second, to tackle complex uncertainties, prescribed-time-driven adaptive update and disturbance estimation laws are separately formulated for precise parameter learning and active disturbance feedforward compensation. Furthermore, a smooth nonlinear robust term is specifically integrated to suppress the residual errors induced by parameter adaptation. Based on the transformed system, a novel control framework integrating error constraints, adaptive parameter estimation, and prescribed-time performance is developed. Rigorous Lyapunov stability analysis proves that the proposed controller not only strictly prevents the system output from violating the constraint boundaries throughout the entire operation, but also ensures that the tracking error converges rapidly and smoothly to a small bounded region near the origin within a time that can be independently predetermined by the designer. Finally, the effectiveness and superiority of the proposed control strategy are fully validated through simulations.

To this end, there is still little research on the staycation travel intentions of potential tourists in the context of social media. For this study, a quantitative study design is utilized to better understand how social media information influences potential tourists' staycation travel intentions in terms of desire, trends/fads, and experience. Using the S-O-R theory, this study generates a novel framework that provides a more in-depth view of the dynamic interactions between social media information and staycation destination choices. The study empirically validates, based on 1250 potential tourists, that the influence of social media information (e.g., value, self-congruence, preference), particularly its reliability, on travel intention is established by mediating the effects via cognitive image. The findings are useful to fill gaps in theory underlining staycation tourists’ behavior and provide actionable recommendations to industry stakeholders. In addition, this study offers practical insights for both tourism practitioners and policymakers on harnessing social media as a strategic tool for staycation destination marketing and management. By exploring these complex dynamics through an interdisciplinary approach, this study makes a significant contribution to the academic discourse on staycation tourism marketing and consumer behavior in the digital age.

In recent years, UAV-based multispectral object detection has demonstrated tremendous potential in smart city traffic management and disaster response capabilities. However, most existing methods focus on better aligning or fusing the two images, while neglecting the blurring of object edges in infrared images under adverse conditions, making it more challenging to distinguish between the foreground and background, thus increasing the difficulty of object detection.To address this issue, we propose a novel cross-modal edge-enhanced detector for UAV-based multispectral object detection.First, we design an Edge Feature Enhancement Module that uses differential convolution to compute the difference between the weighted fusion pooling layer and the input feature map, thereby enhancing edge features in the images.Next, we design a Multi-Scale Feature Fusion Module that employs dilated convolution to expand the receptive field by inserting gaps between kernel elements without increasing the kernel size.This enables the model to detect objects of varying sizes and adapt to resolution changes caused by UAV flight dynamics. Finally, we introduce a Cross-Modal Feature Fusion Module that leverages a self-attention mechanism to learn, adjust, and fuse complementary information from both modalities, enhancing the model, s robustness and improving feature representation across the two spectra.CMEE-Det outperforms existing methods on the DroneVehicle dataset and two other multimodal object detection datasets.

In order to determine the elements geochemical combination characteristics of the Bayan Obo deposit,the ore samples of Bayan Obo main ore body were systematically collected,and 36 indicators including major and trace elements were determined.Firstly,the original geochemical data were transformed using the centered log-ratio(clr)transformation method,and then the transformed data are used to divided the elements combination characteristics by cluster and factor analysis.The results showed that through cluster analysis,the 36 elemental indicators were aggregated into 4 elements combination clusters and 1 single-element cluster.Through factor analysis,the distribution characteristics of 6 representative factors can represent the distribution characteristics of the 36 elemental indicators.By comparing the results of cluster and factor analysis,the con-clusions of both methods were basically consistent.This study explains the symbiotic combination characteristics of elements in Bayan Obo deposit and their combination differences from a mineralogical perspective,and also provided reference basis for in-ferring the occurrence and mineral genesis of related elements.