China Railway Corporation

As deep learning models are increasingly embedded as critical components within complex socio-technical systems, understanding and evaluating their systemic robustness against adversarial perturbations has become a fundamental concern for system safety and reliability. Deep neural networks (DNNs) are highly effective in visual recognition tasks but remain vulnerable to adversarial perturbations, which can compromise their reliability in safety-critical applications. Existing attack methods often distribute perturbations uniformly across the input, ignoring the spatial heterogeneity of model sensitivity. In this work, we propose the Spatially Distributed Perturbation Strategy with Smoothed Gradient Sign Method (SD-SGSM), a adversarial attack framework that exploits decision-dependent regions to maximize attack effectiveness while minimizing perceptual distortion. SD-SGSM integrates three key components: (i) decision-dependent domain identification to localize critical features using a deterministic zero-out operator; (ii) spatially adaptive perturbation allocation to concentrate attack energy on sensitive regions while constraining background disturbance; and (iii) gradient smoothing via a hyperbolic tangent transformation to enable fine-grained and continuous perturbation updates. Extensive experiments on CIFAR-10 demonstrate that SD-SGSM achieves near-perfect attack success rates (ASR 99.9%) while substantially reducing ℓ2 distortion and preserving high structural similarity (SSIM 0.947), outperforming both single-step and momentum-based iterative attacks. Ablation studies further confirm that spatial distribution and gradient smoothing act as complementary mechanisms, jointly enhancing attack potency and visual fidelity. These findings underscore the importance of spatially aware, decision-dependent adversarial strategies for system-level robustness assessment and the secure design of AI-enabled systems.

China government sends students from ethnic regions to study in the eastern developed regions. However, few studies have focused on whether students from ethnic regions can adapt to new cultural environments. This study aimed to investigate the mediating roles of teacher support and peer relations on the relationship between cultural intelligence and internalizing problems among ethnic minority students in China. A three-wave lagged study was conducted from October 2023 to April 2024. 520 ethnic minority students were recruited. This study found that cultural intelligence was positively correlated with teacher support and peer relations, and negatively correlated with internalizing problems. Teacher support was positively correlated with peer relations and negatively correlated with internalizing problems. Peer relations was negatively correlated with internalizing problems. Teacher support and peer relations played parallel mediating roles in the relationship between cultural intelligence and internalizing problems. Cultural intelligence, as an individual internal asset, was not only negatively associated with internalizing problems of ethnic minority students, but also helped them to establish good interpersonal relationships with teachers and peer. Teacher support and positive peer relations as external assets were both protective factors for internalizing problems among ethnic minority students.

The coupler-type formwork support system is extensively adopted in China’s construction industry owing to its structural efficiency and cost-effectiveness. However, quantitative design methods for such systems remain challenging due to computational inaccuracies in characterizing semi-rigid coupler joints and imperfections in practical implementations. This study investigates the load-carrying capacity of scaffold structures with inherent defects through advanced numerical simulations. A hybrid FE modelling approach was developed, integrating 3D shell elements for critical coupler regions and beam elements for structural members. The model incorporated semi-rigid joint behaviour for right-angle couplers, with experimental validation confirming its accuracy. Parametric analyses evaluated three key factors, i.e. steel grade, coupler casting material properties, and steel tube wall thickness. Results demonstrate that steel strength and tube thickness significantly enhance structural performance, Q450 steel increased ultimate load capacity by 34.35% versus Q235, while 3.6 mm wall thickness improved capacity by 23.78% compared to 3.0 mm. Crucially, coupler material properties exhibited negligible impact on system-level behaviour.

Unsupervised domain adaptation (UDA) is critical for cross-modal 3D semantic segmentation in real scenarios, as it significantly reduces the annotation costs in new domains. Existing methods primarily focus on various alignment across domains or modalities, however, domain shift and modal differences are inbuilt, and learning under the guidance of source semantic knowledge instead of forced alignment may be more favorable for UDA tasks. Motivated by this insight, we propose cross Domain Label-Guided learning (xDLG) for cross-modal UDA 3D semantic segmentation, which utilizes multimodal pseudo-labels generated online as a bridge for cross-domain label knowledge transfer. Specifically, we propose two types of label-guided learning: Label-Guided Semantic Prototype Learning (LG-SPL) and Label-Guided Pseudo-Label Learning (LG-PLL). First, in LG-SPL, we utilize massive class instances from the source domain to correct the semantic prototypes in the target domain, embedding genuine semantic similarity to tackle the issue of class confusion. Next in LG-PLL, we conceptualize the refinement of pseudo-labels as a reverse diffusion process, leveraging label information from the source domain as a guiding signal to adjust the semantic distribution of pseudo-labels on a global scale. Furthermore, we incorporate domain-specific cues from target 2D images as supplementary conditions throughout the denoising stages, ensuring semantic consistency at finer details. We evaluate our method on 3D semantic segmentation tasks. The results demonstrate that the proposed method achieves state-of-the-art performance across multiple cross-modal UDA scenarios.