Gradient descent learning in and out of equilibrium

Relations between the off thermal equilibrium dynamical process of on-line learning and the thermally equilibrated off-line learning are studied for potential gradient descent learning. The approach of Opper to study on-line Bayesian algorithms is used for potential based or maximum likelihood learning. We look at the on-line learning algorithm that best approximates the off-line algorithm in the sense of least Kullback-Leibler information loss. The closest on-line algorithm works by updating the weights along the gradient of an effective potential, which is different from the parent off-line potential. A few examples are analyzed and the origin of the potential annealing is discussed