Training prohibited item detection models requires a large amount of X-ray security images, but collecting and annotating these images is time-consuming and laborious. To address data insufficiency, X-ray security image synthesis methods composite images to scale up datasets. However, previous methods primarily follow a two-stage pipeline, where they implement labor-intensive foreground extraction in the first stage and then composite images in the second stage. Such a pipeline introduces inevitable extra labor cost and is not efficient. In this paper, we propose a one-stage X-ray security image synthesis pipeline (Xsyn) based on text-to-image generation, which incorporates two effective strategies to improve the usability of synthetic images. The Cross-Attention Refinement (CAR) strategy leverages the cross-attention map from the diffusion model to refine the bounding box annotation. The Background Occlusion Modeling (BOM) strategy explicitly models background occlusion in the latent space to enhance imaging complexity. To the best of our knowledge, compared with previous methods, Xsyn is the first to achieve high-quality X-ray security image synthesis without extra labor cost. Experiments demonstrate that our method outperforms all previous methods with 1.2% mAP improvement, and the synthetic images generated by our method are beneficial to improve prohibited item detection performance across various X-ray security datasets and detectors. Code is available at https://github.com/pILLOW-1/Xsyn/.

The inherent autocorrelation of time series data presents an ongoing challenge to multivariate time series prediction. Recently, a widely adopted approach has been the incorporation of frequency domain information to assist in longterm prediction tasks. Many researchers have independently observed the spectral bias phenomenon in neural networks, where models tend to fit low-frequency signals before high-frequency ones. However, these observations have often been attributed to the specific architectures designed by the researchers, rather than recognizing the phenomenon as a universal characteristic across models. To unify the understanding of the spectral bias phenomenon in long-term time series prediction, we conducted extensive empirical experiments to measure spectral bias in existing mainstream models. Our findings reveal that virtually all models exhibit this phenomenon. To mitigate the impact of spectral bias, we propose the FreLE (Frequency Loss Enhancement) algorithm, which enhances model generalization through both explicit and implicit frequency regularization. This is a plug-and-play model loss function unit. A large number of experiments have proven the superior performance of FreLE. Code is available at https://github.com/Chenxing-Xuan/FreLE.

The emergence of heterogeneous data brings new challenges to Federated Learning (FL). Unlike homogeneous datasets, heterogeneous data are inherently characterized by sample domain center offsets, which makes traditional federated learning often suffer from poor model results due to catastrophic forgetting and increased communication costs when dealing with such data. To address this problem, we introduce FedNHN, a novel Personalized Federated Learning (PFL) algorithm. FedNHN regularizes the local training loss by quantifying the disparity between the local and global models to generate pseudo-data feature representations. Specifically, it generates two types of pseudo-data: Highly Pseudo and Perturbed Data, by employing the Fast Gradient Sign Method (FGSM) with both the global model and local data. Furthermore, FedNHN leverages the Hilbert-Schmidt Independence Criterion (DN-HSIC) to assess the distances between the feature representations of the generated data by the global model and the local model, respectively, and utilizes the Centered Kernel Alignment (CKA) similarity based on DN-HSIC to regulate the local training loss effectively. Our mathematical analyses show that our design can maintain the utility of the federated model under heterogeneous data, and extensive evaluations on three classical datasets illustrate the effectiveness and performance of our proposed implementation.