Xiaoyu Xie
Senior Research Scientist at Meta · Ph.D. at Northwestern University
My research integrates physics and prior knowledge into machine learning to build interpretable, scalable AI, with applications in scientific discovery, additive manufacturing, fluid mechanics, and ads bidding. My Ph.D. in Mechanical Engineering was advised by Prof. Wing Kam Liu.
Research
01
Discovery of Governing Laws
Dimensionless numbers, scaling laws, and PDEs from scarce, noisy data.
02Reduced-Order Modeling
Compact latent dynamics for fast, accurate spatiotemporal surrogates.
03ML-Based Digital Twins
Process–structure–property prediction in metal additive manufacturing.
Also interested in computer vision and multimodal learning.
News
- Nov 2026Our paper "UNIDE: Cross-PDE Learning for Noise-Robust Discovery of Governing Equations" was released on arXiv!
- Dec 2025Our paper "A Tutorial on Dimensionless Learning: Geometric Interpretation and the Effect of Noise" for PyDimension 2.0 was released on arXiv!
- Oct 2025Our paper "Unifying machine learning and interpolation theory via interpolating neural networks" was accepted by Nature Communications!
- Sep 2024I joined Meta as a Research Scientist.
- Jun 2024Completed a three-month internship at Amazon Annapurna Labs, focusing on optimizing distributed inference for multimodal deep learning models.
- Feb 2024Published internship paper on operator learning: "Smooth and Sparse Latent Dynamics in Operator Learning with Jerk Regularization" on arXiv.
Older news
- Jun 2023Completed a three-month Research Internship at Mitsubishi Electric Research Laboratories (MERL).
- Mar 2023Invited to give a talk on Dimensionless Learning at the CMU SciML Webinar. (Recording)
- Dec 2022Our paper "Data-driven discovery of dimensionless numbers and governing laws from scarce measurements" was published in Nature Communications!
- Aug 2022Our team won three 1st places and two 2nd places in the AM-Bench Challenge at NIST, US!
- Jun 2021Our work was published in npj Computational Materials (Nature Partner Journal)!
Honors & Awards
- 2022AM-Bench 2022 Challenge Awards, NIST — three 1st places, one 2nd place (scan geometry, solid cooling rate, melt pool geometry; time above melting)
- 2021Predictive Science and Engineering Design Fellowship, Northwestern University
- 2021Conference Award, MMLDT-CSET 2021
- 2021Conference Award, 16th U.S. National Congress on Computational Mechanics (USNCCM16)
- 2020Walter P. Murphy Fellowship, Northwestern University
Education
- 2020–2025Ph.D., Mechanical Engineering, Northwestern University · Advisor: Prof. Wing Kam Liu
- 2014–2017M.S., General and Fundamental Mechanics, University of Chinese Academy of Sciences · Institute of Mechanics