Statistical Mechanics of Linear and Nonlinear Time-Domain Ensemble Learning

TL;DR

This paper introduces time-domain ensemble learning, analyzing its effectiveness for linear and nonlinear models, and demonstrates its potential to significantly improve generalization performance over traditional space-domain ensemble methods.

Contribution

It presents a novel time-domain ensemble learning approach and provides analytical and numerical evidence of its advantages for both linear and nonlinear models.

Findings

Time-domain ensemble learning is twice as effective as conventional methods for linear models.

Nonlinear models exhibit nonmonotonic generalization error behaviors at small learning rates.

Performance improvements are achieved by leveraging divergence of students in the time domain.

Abstract

Conventional ensemble learning combines students in the space domain. In this paper, however, we combine students in the time domain and call it time-domain ensemble learning. We analyze, compare, and discuss the generalization performances regarding time-domain ensemble learning of both a linear model and a nonlinear model. Analyzing in the framework of online learning using a statistical mechanical method, we show the qualitatively different behaviors between the two models. In a linear model, the dynamical behaviors of the generalization error are monotonic. We analytically show that time-domain ensemble learning is twice as effective as conventional ensemble learning. Furthermore, the generalization error of a nonlinear model features nonmonotonic dynamical behaviors when the learning rate is small. We numerically show that the generalization performance can be improved remarkably by using this phenomenon and the divergence of students in the time domain.