We discuss the on-line learning of probability distributions in a reparametrization covariant formulation. Reparametrization covariance plays an essential role not only to respect an intrinsic property of “information” but also for pattern recognition problems. We can obtain an optimal on-line learning algorithm with reparametrization invariance, where the conformal gauge connects a covariant formulation with a noncovariant one in a natural way.
|Number of pages||1|
|Journal||Physical Review E - Statistical Physics, Plasmas, Fluids, and Related Interdisciplinary Topics|
|Publication status||Published - Jan 1 2001|
ASJC Scopus subject areas
- Statistical and Nonlinear Physics
- Statistics and Probability
- Condensed Matter Physics