I am a postdoctoral researcher in the Department of Statistics at Carnegie Mellon University, working with Aaditya Ramdas and Alessandro Rinaldo. I recently graduated from the same department with a Ph.D. in Statistics.

Research

My research interest lies in understanding the sequential and adaptive nature of data analysis. I study how commonly used statistical inference procedures behave under the presence of an analyst’s data-dependent choices. My current projects focus on designing and analyzing nonasymptotic sequential testings and online change-point detection procedures.

During the Ph.D., my thesis’s focus was to qualitatively and quantitatively measure the selection bias caused by different sorts of human interventions in settings of adaptive experimentation like multi-armed bandits, advised by Aaditya Ramdas and Alessandro Rinaldo. I have also worked extensively with Larry Wasserman to understand the geometrical and topological structure of data-generating processes.

Keywords: Anytime-valid inference, Sequential test, Multi-armed bandit, Change-point detection

Papers

Nonparametric iterated-logarithm extensions of the sequential generalized likelihood ratio test
J. Shin, A. Ramdas, A. Rinaldo
Submitted to IEEE Journal on Selected Areas in Information Theory arXiv, code

On conditional versus marginal bias in multi-armed bandits
J. Shin, A. Ramdas, A. Rinaldo
Thirty-seventh International Conference on Machine Learning (ICML 2020) arXiv

Are sample means in multi-armed bandits positively or negatively biased?
J. Shin, A. Ramdas, A. Rinaldo
Neural Information Processing Systems (NeurIPS 2019, Spotlight) arXiv

On the bias, risk and consistency of sample means in multi-armed bandits
J. Shin, A. Ramdas, A. Rinaldo
Submitted to SIAM Journal on Mathematics of Data Science arXiv

Predictive clustering
J. Shin, A. Rinaldo, L. Wasserman
Submitted to Electronic Journal of Statistics arXiv

Uniform Convergence Rate of the Kernel Density Estimator Adaptive to Intrinsic Dimension.
J. Kim, J. Shin, A. Rinaldo, L. Wasserman
Thirty-sixth International Conference on Machine Learning (ICML 2019) arXiv

Homotopy Reconstruction via the Cech Complex and the Rips Complex
J. Kim, J. Shin, F. Chazal, A. Rinaldo, L. Wasserman
Symposium on Computational Geometry (SoCG 2020) arXiv

Persistent homology of KDE filtration on Rips complex
J. Shin, J. Kim, A. Rinaldo, L. Wasserman
In preparation.

Contact

Jaehyeok Shin Ph.D. in Statistics
Department of Statistics
Carnegie Mellon University

Last updated: 2020-10-28.