Shihao Wu 

About Me

Hi! I am a third-year PhD student in Statistics at the University of Michigan, jointly supervised by Professor Ji Zhu and Professor Gongjun Xu. Prior to that, I received my Bachelor's degree in Data Science from Fudan University in 2021. 

Research Interests

I am interested in and working on statistical problems that emerge from the need to analyze large-scale, complex-structured datasets in the big data era.  Specifically,

• Statistical embedding:  
  Although humans are able to collect an rapidly increasing amount of data, the essential information in these datasets remains deeply buried due to the complex structures of the datasets. Observed data for the subjects of interest may be noisy, non-interpretable, and even non-numerical. I think about how to leverage statistics and machine learning tools to provide the data with de-noised, interpretable and ready-to-use embeddings, which not only reveal insights about the data but are also useful for a series of downstream tasks. 

• Domain-dependent and flexible uncertainty quantification:  
  With the explosion of data resources, there have been so many powerful algorithms (many of which are black-box) giving us the desired outputs: embeddings, predictions, estimates of model parameters, and so on. With a statistics background, I always ask myself this question: to what extent can we trust these outputs, and what are the practically meaningful measures of "extent" under different circumstances? Moreover, inference procedures typically need to enforce assumptions about the data-generating mechanism. Some of these assumptions are strong and hard to verify in practice. I am interested in developing inference algorithms that require weak and verifiable assumptions as possible.

Education

• Ph.D. in Statistics, University of Michigan, 2021 - present

• Sc.B. in Data Science, Fudan University, 2017 - 2021

Selected Awards and Honors

• 2024 IMS Hannan Graduate Student Travel Award, IMS

• 2024 Student Paper Award, Statistical Learning and Data Science Section, ASA
  ''A general latent embedding approach for modeling high-dimensional hyperlinks'' with Gongjun Xu and Ji Zhu

• 2022 Best Poster Presentation Award, UMichSML
  ''Supervised homogeneity fusion: a combinatorial approach'' with Wen Wang, Ziwei Zhu, Ling Zhou and Peter X.K. Song

• 2021-2022 Outstanding First Year Ph.D. Student, Department of Statistics, University of Michigan