About me

About me

I am a 4th-year Ph.D. Student in the Department of Computer Science at the University of Maryland, College Park (UMD). As a research assistant at UMD, I work with my Ph.D. advisor Dr.Tom Goldstein. My research goal is to develop practical machine learning/deep learning (ML/DL) models that can generalize to real-world data. Relevant research directions include robustness to distribution shift, out-of-distribution generalization, etc. I am interested in studying these problems on different applications, including computer vision and multimodal learning.


  • Test-Time Prompt Tuning for Zero-Shot Generalization in Vision-Language Models [paper] [code]

    M. Shu , W. Nie, DA. Huang, Z. Yu, T. Goldstein, A. Anandkumar, and C. Xiao, NeurIPS, 2022

  • Where do models go wrong? Parameter-space saliency maps for explainability. [paper]

    R. Levin*, M. Shu*, E. Borgnia*, F. Huang, M. Goldblum, T. Goldstein, NeurIPS, 2022

  • The Close Relationship between Contrastive Learning and Meta Learning [paper] [code]

    R. Ni*, M. Shu*, H. Souri, M. Goldblum, and T. Goldstein, ICLR, 2022

  • Encoding Robustness to Image Style via Adversarial Feature Perturbations [paper] [code]

    M. Shu , Z. Wu, M. Goldblum, and T. Goldstein, NeurIPS, 2021

  • Gradient-Free Adversarial Training against Image Corruption for Learning-based Steering [paper] [code]

    Y. Shen, L. Zheng, M. Shu , W. Li, T. Goldstein and M. Lin, NeurIPS, 2021

  • Adversarial Differentiable Data Augmentation for Autonomous Systems [paper] [code]

    M. Shu , Y. Shen, M. Lin, and T. Goldstein, ICRA, 2021

  • Towards Accurate Quantization and Pruning via Data-free Knowledge Transfer. [paper]

    C. Zhu, Z. Xu, A. Shafahi, M. Shu , A. Ghiasi, and T. Goldstein, SNN (Workshop), 2021

  • Headless Horseman: Adversarial Attacks on Transfer Learning Models. [paper]

    A. Abdelkader, M. Curry, L. Fowl, T. Goldstein, A. Schwarzschild, M. Shu , C. Studer, C. Zhu, ICASSP, 2020