Timezone: »

Data-Dependence of Plateau Phenomenon in Learning with Neural Network --- Statistical Mechanical Analysis
Yuki Yoshida · Masato Okada

Wed Dec 11 05:00 PM -- 07:00 PM (PST) @ East Exhibition Hall B + C #237

The plateau phenomenon, wherein the loss value stops decreasing during the process of learning, has been reported by various researchers. The phenomenon is actively inspected in the 1990s and found to be due to the fundamental hierarchical structure of neural network models. Then the phenomenon has been thought as inevitable. However, the phenomenon seldom occurs in the context of recent deep learning. There is a gap between theory and reality. In this paper, using statistical mechanical formulation, we clarified the relationship between the plateau phenomenon and the statistical property of the data learned. It is shown that the data whose covariance has small and dispersed eigenvalues tend to make the plateau phenomenon inconspicuous.

Author Information

Yuki Yoshida (The University of Tokyo)
Masato Okada (The University of Tokyo)