Deep learning-based face recognition models are vulnerable to adversarial attacks. In contrast to general noises, the presence of imperceptible adversarial noises can lead to catastrophic errors in deep face recognition models. The primary difference between adversarial noise and general noise lies in its specificity. Adversarial attack methods give rise to noises tailored to the characteristics of the individual image and recognition model at hand. Diverse samples and recognition models can engender specific adversarial noise patterns, which pose significant challenges for adversarial defense. Addressing this challenge in the realm of face recognition presents a more formidable endeavor due to the inherent nature of face recognition as an open set task. In order to tackle this challenge, it is imperative to employ customized processing for each individual input sample. Drawing inspiration from the biological immune system, which can identify and respond to various threats, this paper aims to create an artificial immune system to provide adversarial defense for face recognition. The proposed defense model incorporates the principles of antibody cloning, mutation, selection, and memory mechanisms to generate a distinct “antibody” for each input sample, wherein the term “antibody” refers to a specialized noise removal manner. Furthermore, we introduce a self-supervised adversarial training mechanism that serves as a simulated rehearsal of immune system invasions. Extensive experimental results demonstrate the efficacy of the proposed method, surpassing state-of-the-art adversarial defense methods. The source code is available here , or you can visit this website: https://github.com/RenMin1991/SIDE
Railway obstacle intrusion detection is crucial for safe train operation, especially as high-speed train continues to develop rapidly. However, existing algorithms still have limitations in detecting distant small objects, adapting to low-light conditions and adverse weather. Moreover, most studies overlook the importance of assessing the risk levels of obstacles. To address these issues, a railway obstacle intrusion detection and risk assessment method based on MSIA-YOLOv8 and DALNet is proposed in this paper. Through comprehensive improvements in network architecture and loss function (Inner-Focal-CIoU), MSIA-YOLOv8 improves the accuracy of multi-scale obstacle detection, especially for distant small objects. The Frequency Domain Aggregation and Enhancement (FDAE) further enhances detection accuracy and robustness in dark and bad-weather conditions while ensuring real-time performance. The algorithm achieves 97.3% mAP with a speed of 137 FPS on the railway dataset. It outperforms YOLOv8 and meets the speed, accuracy and computational requirements of railway scenarios. Furthermore, the integration of DALNet enables the detection of railway obstacle intrusion, as well as the classification and assessment of their risk levels.
Under natural ambient temperatures, the CRTS III slab track undergoes warping deformation and assumes a locally supported state, which intensifies dynamic impacts during train passage. To investigate these complex interactions, this study develops a novel simulation framework. A refined finite element model based on thermalmechanical coupled theory was developed to analyze the time-varying temperature field and associated thermal deformations. Additionally, a vehicle-track real-time collaborative simulation model was established using a multi-time-step explicit-implicit solution method to examine structural vibrations at different spatial locations and the evolution of interlayer contact conditions under coupled temperature and train load. The results indicate that the temperature field within the track structure exhibits significant nonlinear variation and phase lag along the depth. Positive temperature gradients induce periodic rail irregularities and slab warping, reducing global support stiffness by up to 9.05 % and decreasing interlayer contact area by up to 83.5 %. This localized separation concentrates stress at slab corners, posing a risk of damage. The coupling of temperature and train load significantly amplifies structural vibrations and alters their spatial distribution, with resonance observed near 40-50 Hz. As train speed increases, the characteristic frequency associated with the slab length approaches and amplifies the wheel-rail coupling resonance. The findings present a novel approach to modeling the coupling of thermal effects with vehicle-track interaction dynamics and provide theoretical support for the design and maintenance of slab tracks under high-speed operating conditions.
Fault detection of railway fasteners is vital for ensuring train operation safety within intelligent rail maintenance systems. However, existing Transformer-based detectors often incur high computational overhead and exhibit limited adaptability to multi-scale variations and complex background interference. To address these limitations, we propose SaM-DETR, a novel object detection framework that deeply couples causal saliency enhancement with a Mamba-based Multi-scale State Space Module (MSSM). Specifically, we introduce a differentiable Top-K selection mechanism combined with a causal invariance loss to adaptively learn the optimal foreground-background ratio, effectively suppressing spurious correlations and enhancing salient feature extraction. Furthermore, we replace the vanilla transformer encoder with a linear-complexity Mamba-based MSSM encoder, enabling efficient long-range feature fusion while dramatically reducing giga floating-point operations (GFLOPs). Extensive experiments on two real-world railway fastener datasets with different resolutions demonstrate that SaM-DETR achieves a balanced trade-off among detection accuracy, computational efficiency, and inference speed, outperforming mainstream DEtection TRansformer (DETR)-based models. These results validate the robustness and lightweight advantages of its novel causal-saliency-driven multi-scale design in multi-resolution industrial scenarios.
Hydrogen storage remains a critical bottleneck for clean energy transition, where metal-organic frameworks (MOFs) with ultrahigh surface areas and tunable pore environments offer unprecedented opportunities. Herein, we systematically engineered the pore chemistry of UiO-66 via functional group substitution to elucidate the structure-performance relationship for hydrogen adsorption. Seven functionalized UiO-66-X (X = H, CH3, NH2, OH, F, Br, NO2) derivatives were synthesized using a solvothermal approach, enabling precise control over electronic properties through modification of the 1,4-benzenedicarboxylate (BDC) linker. Under cryogenic conditions (77 K, 50 bar), all functionalized variants exhibited reduced hydrogen uptake compared to the parent UiO-66-H (2.11 wt%), while at ambient temperature (298 K, 100 bar), only UiO-66-F (0.27 wt%) surpassed the storage capacity of UiO-66-H (0.22 wt%). Density functional theory (DFT) calculations revealed that electron-donating groups (e.g., CH3, -NH2, -OH) enhance H2 adsorption via polarization-induced dipole interactions, while electron-withdrawing substituents (e.g., -F,-Br, -NO2) promote charge-transfer-mediated binding. This indicates that the electronic effect of functional groups is one of the dominant factors influencing hydrogen storage performance. These findings establish a correlation between functional group electron affinity and H2 binding energy, providing a rational design strategy for high-performance MOF-based hydrogen storage materials.
