John Y. Shin is a Senior Research Associate at Numerati® Partners (and Enumeration/Evaluation lead), focusing on classification and error rate quantification in real-time machine learning systems, which leverage physical principles.
He graduated summa cum laude with a B.Sc. in Physics from the University of California, Santa Cruz, where he first-authored a publication on frustrated magnetic systems, utilizing techniques from ab-initio quantum chemistry. He has also published work on image deconvolution techniques applied to nanoscale imaging used in material science for understanding superconducting pairing mechanisms in unconventional superconductors.
Transitioning from Computational Physics, John recently received a Master’s degree in Computer Science from New York University. His current research and interests are in the area of robust machine learning, where he has worked on utilizing Langevin dynamics to improve the robustness and calibration of machine learning models, as well as theoretical work on the generalization gap of deep learning models.