Efficient and principled score estimation

Abstract

We propose a fast method with statistical guarantees for learning an exponential family density model where the natural parameter is in a reproducing kernel Hilbert space, and may be infinite-dimensional. The model is learned by fitting the derivative of the log density, the score, thus avoiding the need to compute a normalization constant. Our approach… (More)

3 Figures and Tables

Topics

  • Presentations referencing similar topics