Haiwen Huang

Haiwen Huang

PhD student at University of Tübingen, interested in the intersection of machine learning and computer vision

About me

I am Haiwen Huang, my Chinese name is 黄海文. I am a PhD student at University of Tübingen, co-supervised by Prof. Andreas Geiger and Dr. Dan Zhang, starting in Jan. 2022. I am also an ELLIS PhD student.

My career goal is to empower industry with AI to boost the producitivity of human society. To achieve this, we need AI systems to be trustworthy and powerful in various environments.

  • To be trustworthy, the AI systems need to be able to tell when they are unreliable (OOD detection);
  • To be powerful, the AI systems need to be robust to the (unseen) changes of the environments and domains (OOD generalization).

In order to achieve generalization to unseen objects and environments, a model requires additional information that is invariant across categories and domains. My PhD studies aim to explore methods for incorporating knowledge from pretrained large models to achieve out-of-distribution generalization in various tasks.

I believe that using large pretrained models (e.g. Omnidata models and Stable Diffusion) will significantly reduce the cost of AI development and deployment while improving the overall well-being of the world.

Before my PhD studies, I completed my MSc in CS at University of Oxford (supervisor: Prof. Yarin Gal) and my BSc in Math at Peking University (supervisor: Prof. Bin Dong). I have also worked as a researcher for a year at Megvii (previously known as Face++), studying large-scale annotation and OoD detection methods.

My CV is here.

Education

  • University of Tübingen, Germany (Jan. 2022 - Now)
    • Ph.D, Computer Science
    • Advisor: Andreas Geiger and Dan Zhang
  • University of Oxford, UK (Oct. 2020 - Sept. 2021)
    • M.S., Computer Science (Distinction)
    • Advisor: Yarin Gal
  • Peking University, China (Sept. 2015 - June 2019)
    • B.S., Computing Mathematics

Work experience

  • Megvii Technology Inc, Beijing, China, (Feb. 2019 - Sept. 2020)
    • ML Researcher, Data Research Group
    • Research projects: Large-scale data annotation, out-of-distribution detection, few-shot learning
    • Group leader: Xinyu Zhou

Highlighted Research

Teaching