Convex regression in multidimensions

Convex regression in multidimensions

Neyman Seminar
Feb 17, 2021, 04:00 PM - 05:00 PM | On Zoom Evans Hall | Happening As Scheduled
Aditya Guntuboyina, UC Berkeley

Convex regression refers to the problem of estimating an unknown convex function from noisy measurements. The least squares estimator presents a natural estimation strategy in this problem. I will present some recent results on the accuracy of the least squares estimator. An interesting feature of our results is that the least squares estimator is provably suboptimal in some settings of the...