On the optimality of the aggregate with exponential weights for low temperatures

TL;DR

This paper investigates the aggregate with exponential weights (AEW) in regression, revealing its suboptimality at low temperatures unless a Bernstein condition is met, where it becomes optimal with a refined complexity measure.

Contribution

The paper characterizes the conditions under which AEW is optimal or suboptimal in low-temperature regimes, introducing a Bernstein condition for optimality.

Findings

AEW is suboptimal in expectation for low temperatures.

AEW may concentrate around suboptimal functions with high probability.

Under Bernstein condition, AEW achieves optimality in low-temperature regimes.

Abstract

Given a finite class of functions F, the problem of aggregation is to construct a procedure with a risk as close as possible to the risk of the best element in the class. A classical procedure (PAC-Bayesian statistical learning theory (2004) Paris 6, Statistical Learning Theory and Stochastic Optimization (2001) Springer, Ann. Statist. 28 (2000) 75-87) is the aggregate with exponential weights (AEW), defined by \[\tilde{f}^{\mathrm{AEW}}=\sum_{f\in F}\hat{\theta}(f)f,\qquad where \hat{\theta}(f)=\frac{\exp(-({n}/{T})R_n(f))}{\sum_{g\in F}\exp(-({n}/{T})R_n(g))},\] where $T>0$ is called the temperature parameter and $R_n(\cdot)$ is an empirical risk. In this article, we study the optimality of the AEW in the regression model with random design and in the low-temperature regime. We prove three properties of AEW. First, we show that AEW is a suboptimal aggregation procedure in expectation with respect to the quadratic risk when $T\leq c_1$, where $c_1$ is an absolute positive constant (the low-temperature regime), and that it is suboptimal in probability even for high temperatures. Second, we show that as the cardinality of the dictionary grows, the behavior of AEW might deteriorate, namely, that in the low-temperature regime it might concentrate with high probability around elements in the dictionary with risk greater than the risk of the best function in the dictionary by at least an order of $1/\sqrt{n}$. Third, we prove that if a geometric condition on the dictionary (the so-called "Bernstein condition) is assumed, then AEW is indeed optimal both in high probability and in expectation in the low-temperature regime. Moreover, under that assumption, the complexity term is essentially the logarithm of the cardinality of the set of "almost minimizers" rather than the logarithm of the cardinality of the entire dictionary. This result holds for small values of the temperature parameter, thus complementing an analogous result for high temperatures.