On the inability of Gaussian process regression to optimally learn compositional functions

05/16/2022
by   Matteo Giordano, et al.
0

We rigorously prove that deep Gaussian process priors can outperform Gaussian process priors if the target function has a compositional structure. To this end, we study information-theoretic lower bounds for posterior contraction rates for Gaussian process regression in a continuous regression model. We show that if the true function is a generalized additive function, then the posterior based on any mean-zero Gaussian process can only recover the truth at a rate that is strictly slower than the minimax rate by a factor that is polynomially suboptimal in the sample size n.

READ FULL TEXT

Please sign up or login with your details

Forgot password? Click here to reset