Luo Lab（罗倚斯课题组）依托西安交通大学数学与统计学院开展研究，关注张量方法、连续表示与 AI for Science 的交叉问题。我们希望为高维复杂科学数据建立既有结构归纳偏置、又具表达能力的学习模型。
Luo Lab is based in the School of Mathematics and Statistics at Xi’an Jiaotong University. We work at the intersection of tensor methods, continuous representations, and AI for Science, with the goal of building learning models that combine structural inductive bias with strong expressive power for complex scientific data.
实验室当前的核心主题包括 低秩张量建模、连续表示与隐式神经表示、以及 AI4Sci 中的恢复、反演与建模问题。我们从张量分解、函数表示、神经算子和优化机制的角度研究多维数据恢复、尺度连续建模与结构化学习。
Our current research themes include low-rank tensor modeling, continuous and implicit neural representations, and recovery, inversion, and modeling problems in AI4Sci. We study multidimensional data recovery, scale-continuous modeling, and structured learning through tensor decomposition, functional representations, neural operators, and optimization mechanisms.
这些方法被应用于图像修复、去雨、地震去噪、全波形反演、空间转录组解析等任务。我们尤其关注如何把张量结构、连续域建模与深度网络连接起来，形成适合科学数据的统一建模框架。
These methods are applied to image restoration, deraining, seismic denoising, full-waveform inversion, and spatial transcriptomics. A central goal is to connect tensor structure, continuous-domain modeling, and deep networks into a unified framework tailored to scientific data.
我们既重视理论，也重视实际任务中的可用性。实验室关注近似能力、结构先验、优化行为与泛化特性，同时坚持在真实科学场景中验证方法，而不是只停留在抽象 benchmark 上。
We care about both theory and practical utility. The lab studies approximation power, structural priors, optimization behavior, and generalization, while insisting on validation in real scientific tasks rather than only abstract benchmarks.