Hello! I am a pre-doctoral Research Associate at the Machine Learning Department, Carnegie Mellon University.
I am interested in sequential statistics, applied probability and foundations of machine learning. My prior works have focused exclusively on handling heavy-tailed, heterogeneous data in various settings including optimization, control, estimation and beyond.
Bio
- 2023 – now: Research Associate at Carnegie Mellon University (Pittsburgh, PA, USA)
- 2021 – 2022: MS in Machine Learning & Research Assistant at Carnegie Mellon University (Pittsburgh, PA, USA)
- 2020: Research Intern at Vector Institute (Toronto, ON, Canada)
- 2017 – 2021: BEng in Computer Science and Technology at Tsinghua University (Beijing, China)
Publications (see also my Semantic Scholar and Google Scholar pages)
- The Extended Ville’s Inequality for Nonintegrable Nonnegative Supermartingales
- Hongjian Wang, Aaditya Ramdas
- A Unified Recipe for Deriving (time-uniform) PAC-Bayes Bounds
- Ben Chugg, Hongjian Wang, Aaditya Ramdas
- Huber-Robust Confidence Sequences
- Catoni-Style Confidence Sequences for Heavy-tailed Mean Estimation
- Hongjian Wang, Aaditya Ramdas
- To appear in Stochastic Processes and their Applications [GitHub]
- Convergence Rates of Stochastic Gradient Descent under Infinite Noise Variance
- Hongjian Wang, Mert Gürbüzbalaban, Lingjiong Zhu, Umut Şimşekli, Murat A. Erdogdu
- NeurIPS 2021 [Poster]
Misc
A selected list of relevant courses I have taken can be found here.
Contact
hongjian.wang [at] aol.com; hjnwang [at] cmu.edu
I currently live in Shadyside, a neighborhood in Pittsburgh, PA.
ἀριθμὸν ἔξοχον σοφισμάτων
number, the chiefest artifice of all
Aeschylus, Prometheus Bound 459. (Also quoted by André Weil in the foreword to Basic Number Theory)
Hongjian Wang 