Universality laws for Gaussian mixtures in generalized linear models

02/17/2023
โˆ™
by   Yatin Dandi, et al.
โˆ™
1
โˆ™

Let (x_i, y_i)_i=1,โ€ฆ,n denote independent samples from a general mixture distribution โˆ‘_cโˆˆ๐’žฯ_cP_c^x, and consider the hypothesis class of generalized linear models ลท = F(ฮ˜^โŠคx). In this work, we investigate the asymptotic joint statistics of the family of generalized linear estimators (ฮ˜_1, โ€ฆ, ฮ˜_M) obtained either from (a) minimizing an empirical risk Rฬ‚_n(ฮ˜;X,y) or (b) sampling from the associated Gibbs measure exp(-ฮฒ n Rฬ‚_n(ฮ˜;X,y)). Our main contribution is to characterize under which conditions the asymptotic joint statistics of this family depends (on a weak sense) only on the means and covariances of the class conditional features distribution P_c^x. In particular, this allow us to prove the universality of different quantities of interest, such as the training and generalization errors, redeeming a recent line of work in high-dimensional statistics working under the Gaussian mixture hypothesis. Finally, we discuss the applications of our results to different machine learning tasks of interest, such as ensembling and uncertainty

READ FULL TEXT

Please sign up or login with your details

Forgot password? Click here to reset