Shengyi He

About Me

I'm currently an incoming fourth-year PhD student at Columbia IEOR advised by Prof. Henry Lam. Before joing Columbia IEOR, I obtained my B.S. degree in Statistics from the School of Mathematical Science at Peking University. My research topics include error reduction and uncertainty quantification in Monte Carlo methods and stochastic optimization.

Publications

• Deep Probabilistic Accerlerated Evaluation: A robust certifiable rare-event simulation methodology for black-box safety-critical systems , with Mansur Arief, Zhiyuan Huang, Guru Koushik Senthil Kumer, Yuanlu Bai, Wenhao Ding, Henry Lam, and Ding Zhao, Proceedings of The 24th International Conference on Artificial Intelligence and Statistics (AISTATS), 2021.

• Higher-order coverage error analysis for batching and sectioning, with Henry Lam, Proceedings of the Winter Simulation Conference (WSC) , 2021.

  WSC 2021 INFORMS I-SIM PhD Colloquium Award

• Batching on biased estimators, with Henry Lam, to appear in Proceedings of the Winter Simulation Conference (WSC), 2022.

• Importance sampling for rare-event gradient estimation, with Yuanlu Bai, Henry Lam, Guangxin Jiang, and Michael Fu, to appear in Proceedings of the Winter Simulation Conference (WSC), 2022.

• Adaptive importance sampling for efficient stochastic root finding and quantile estimation, with Michael Fu, Guangxin Jiang, and Henry Lam, under review in Operations Research.

• Certifiable deep importance sampling for rare-event simulation of black-box systems, with Mansur Arief, Yuanlu Bai, Wenhao Ding, Zhiyuan Huang, Henry Lam, and Ding Zhao, under review in Operations Research.

• Higher-order expansion and Bartlett correctability of distributionally robust optimization, with Henry Lam, under review in Annals of Statistics.

• Higher-order coverage errors of batching methods via Edgeworth expansions on t-statistics, with Henry Lam, under review in Annals of Statistics.

• A robust approach to assess input distributional uncertainty: Asymptotic approximations and computations, with Henry Lam, in preparation.

• A cheap iterated bootstrap to enhance coverage accuracy, with Henry Lam, in preparation.