Xiaoyi Wen | Searcharxiv

arXiv2026

RKHS-based Latent Position Random Graph Correlation

In this article, we consider the problem of testing whether two latent position random graphs are correlated. We propose a test statistic based on the kernel method and introduce the estimation procedure based on the spectral decomposition of adjacency matrices. Even if no kernel function is specified, the sample graph covariance based on our proposed estimation method will converge to the population version. The asymptotic distribution of the sample covariance can also be obtained. We design a procedure for testing independence under permutation tests and demonstrate that our proposed test statistic is consistent and valid. Our estimation method can be extended to the spectral decomposition of normalized Laplacian matrices and inhomogeneous random graphs. Our method achieves promising results on both simulated and real data.

stat.ME↗

arXiv2026

Distributional Deep Learning for Super-Resolution of 4D Flow MRI under Domain Shift

Super-resolution is widely used in medical imaging to enhance low-quality data, reducing scan time and improving abnormality detection. Conventional super-resolution approaches typically rely on paired datasets of downsampled and original high resolution images, training models to reconstruct high resolution images from their artificially degraded counterparts. However, in real-world clinical settings, low resolution data often arise from acquisition mechanisms that differ significantly from simple downsampling. As a result, these inputs may lie outside the domain of the training data, leading to poor model generalization due to domain shift. To address this limitation, we propose a distributional deep learning framework that improves model robustness and domain generalization. We develop this approch for enhancing the resolution of 4D Flow MRI (4DF). This is a novel imaging modality that captures hemodynamic flow velocity and clinically relevant metrics such as vessel wall stress. These metrics are critical for assessing aneurysm rupture risk. Our model is initially trained on high resolution computational fluid dynamics (CFD) simulations and their downsampled counterparts. It is then fine-tuned on a small, harmonized dataset of paired 4D Flow MRI and CFD samples. We derive the theoretical properties of our distributional estimators and demonstrate that our framework significantly outperforms traditional deep learning approaches through real data applications. This highlights the effectiveness of distributional learning in addressing domain shift and improving super-resolution performance in clinically realistic scenarios.

cs.CV↗