Due to the discrepancy between training and test data, autoencoder-based image fusion methods trained by natural images easily lose vital information of infrared and visible images. Generative adversarial network (GAN)-based methods usually conduct their adversarial learning in image domain, which are difficult to optimize and ultimately affect the image fusion performance. To address these problems, this article proposes an autoencoder-based network, extraction-and-reconstruction network (ERNet), to fuse infrared and visible images. To weaken the discrepancy problem, ERNet’s encoder is trained using infrared and visible images. To stably train the encoder, we conduct adversarial learning in the feature domain of infrared and visible images, which makes the well-trained encoder that could efficiently extract vital features from them. ERNet’s decoder is trained using natural images with a supervised mode, which could reconstruct the final fused image from the vital features extracted by the encoder. The encoder and decoder are alternately trained to further weaken the discrepancy problem. The experimental results show that ERNet can effectively extract vital features of infrared and visible images and obtain higher quality fused images. Our fusion results can be found at https://github.com/suweijian1996/ERNet .
Single-cell RNA sequencing (scRNA-seq) is now a successful technology for identifying cell heterogeneity, revealing new cell subpopulations, and predicting developmental trajectories. A crucial component in scRNA-seq is the precise identification of cell subsets. Although many unsupervised clustering methods have been developed for clustering cell subpopulations, the performance of these methods is prone to be affected by dropout, high dimensionality, and technical noise. Additionally, most existing methods are time-consuming and fail to fully consider the potential correlations between cells. In this paper, we propose a novel unsupervised clustering method called scVGATAE (Single-cell Variational Graph Attention Autoencoder) for scRNA-seq data. This method constructs a reliable cell graph through network denoising, utilizes a novel variational graph autoencoder model integrated with graph attention networks to aggregate neighbor information and learn the distribution of the low-dimensional representations of cells, and adaptively determines the model training iterations for various datasets. Finally, the obtained low-dimensional representations of cells are clustered using kmeans. Experiments on nine public datasets show that scVGATAE outperforms classical and state-of-the-art clustering methods.
This paper studies the dynamics of a network-based SIRS epidemic model with vaccination and a nonmonotone incidence rate. This type of nonlinear incidence can be used to describe the psychological or inhibitory effect from the behavioral change of the susceptible individuals when the number of infective individuals on heterogeneous networks is getting larger. Using the analytical method, epidemic threshold $R_0$ is obtained. When $R_0$ is less than one, we prove the disease-free equilibrium is globally asymptotically stable and the disease dies out, while $R_0$ is greater than one, there exists a unique endemic equilibrium. By constructing a suitable Lyapunov function, we also prove the endemic equilibrium is globally asymptotically stable if the inhibitory factor $\alpha$ is sufficiently large. Numerical experiments are also given to support the theoretical results. It is shown both theoretically and numerically a larger $\alpha$ can accelerate the extinction of the disease and reduce the level of disease.
This study examines heterogeneous energy consumption in Liaoning Province under China’s dual-carbon goals, which operationalize the energy-related Sustainable Development Goals (SDG 7 and SDG 13), extending the sample to 1985–2024. Kernel density estimation characterizes distribution dynamics; a Kaya-based LMDI decomposition quantifies contributions of energy structure, intensity, per-capita output, and population; a semi-parametric time-series model estimates factor associations with a flexible time trend; and a spatial Durbin model on a 30-province panel (2000–2024) assesses spillovers. Consumption grew from 63.25 to 259.85 Mtce (3.69% annually), decelerating markedly after 2012. The LMDI attributes 192% of the increase to per-capita output growth, offset by a −104% intensity effect, with a small net structure effect (+4.9%). In the semi-parametric model, coal share and real GDP are significantly associated with consumption (a semi-elasticity of −0.90 for the coal share and an elasticity of 0.43 for real GDP), and the fitted smooth time component captures unobserved temporal variation not explained by the included covariates. The SDM estimates a modest spatial autoregressive coefficient (ρ = 0.055), positive direct effects of GDP and industrial structure, and significant negative indirect effects of neighbors’ GDP and industrial share, indicating competitive rather than complementary interprovincial energy relationships. Mechanisms such as resource siphoning remain hypotheses for future research. The findings inform sustainable development policy for legacy industrial regions: sustaining efficiency gains, decarbonizing the power supply, and coordinating interprovincial industrial planning.