Self Exploration · Life Exploration · Computer Vision · Artificial Intelligence

Xiaoya Tang 唐晓雅

I study learning-based visual recognition systems, with an emphasis on effective and efficient architectures.

View Research Contact

Gallery

About

Xiaoya is a fourth-year PhD student at the Kahlert School of Computing, University of Utah, affiliated with the Scientific Computing and Imaging Institute (SCI). She is advised by Professor Tolga Tasdizen. She received both her bachelor's degree in Aircraft Design and Engineering and her master's degree in Solid Mechanics from Zhejiang University, Hangzhou, China. Before pursuing her Ph.D., she worked as a finite element analysis (FEA) engineer at CRRC. Moving across mechanics and computer science has sparked her curiosity about how models, humans (mostly herself), and the universe work. Her current work explores how visual models learn from data and how they behave across tasks and settings. Beyond AI, she enjoys philosophy, psychology, exercise, reading, writing, and gardening.

Projects

Selected Projects

Efficient Models

Coming soon: to improve the training and inference efficiency in large multi-modal models.

Representation Learning

Domain Adaptation for Street-View Imagery

Understanding how visual representation models transfer across domains is critical for applying foundation models to real-world datasets. In this work, we present a systematic empirical study of representation learning for Google Street View (GSV) imagery, a domain that differs substantially from standard object-centric benchmarks. We investigate how architecture choice, model scale, self-supervised post-training, and data curation influence downstream performance.

Model Analysis