Learning Curves of Gaussian Process Regression: Approximations and Bounds

摘要

We consider the problem of calculating learning curves (i. e., average gen- eralization performance) of gaussian processes used for regression. On the basis of a simple expression for the generalization error, in terms of the eigenvalue decomposition of the covariance function, we derive a num- ber of approximation schemes. We identify where these become exact and compare with existing bounds on learning curves; the new approxima- tions, which can be used for any input space dimension, generally get substantially closer to the truth.