On risk bounds in isotonic and other shape restricted regression problems

From MaRDI portal



Abstract: We consider the problem of estimating an unknown hetainmathbbRn from noisy observations under the constraint that heta belongs to certain convex polyhedral cones in mathbbRn. Under this setting, we prove bounds for the risk of the least squares estimator (LSE). The obtained risk bound behaves differently depending on the true sequence heta which highlights the adaptive behavior of heta. As special cases of our general result, we derive risk bounds for the LSE in univariate isotonic and convex regression. We study the risk bound in isotonic regression in greater detail: we show that the isotonic LSE converges at a whole range of rates from logn/n (when heta is constant) to n−2/3 (when heta is uniformly increasing in a certain sense). We argue that the bound presents a benchmark for the risk of any estimator in isotonic regression by proving nonasymptotic local minimax lower bounds. We prove an analogue of our bound for model misspecification where the true heta is not necessarily nondecreasing.




Cites work


Cited in
(61)








This page was built for publication: On risk bounds in isotonic and other shape restricted regression problems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2515496)