A Variational Approach to Learning Curves

We combine the replica approach from statistical physics with a variational approach to analyze learning curves analytically. We apply the method to Gaussian process regression. As a main result we derive approximative relations between empirical error measures, the generalization error and the posterior variance.

• Type: Conference
• Year: 2001
• Where: NIPS
• Authors: Dörthe Malzahn, Manfred Opper
• Keywords: Gaussian Process Regression, NIPS 2001, NIPS 2007, Replica Approach, Variational Approach