Robust Sequential Online Prediction with Dynamic Ensemble of Multiple Models: A Concise Introduction

12/04/2021

∙

In this paper, I give a concise introduction to a generic theoretical framework termed Bayesian Dynamic Ensemble of Multiple Models (BDEMM) that is used for robust sequential online prediction. This framework has three major features: (1) it employs a model pool, rather than a single model, to capture possible statistical regularities underlying the data; (2) the model pool consists of multiple weighted candidate models, wherein the model weights are adapted online to capture possible temporal evolutions of the data; (3) the adaptation for the model weights follows Bayesian formalism. These features together define BDEMM. To make this introduction comprehensive, I describe BDEMM from four perspectives, namely the related theories, the different forms of its algorithmic implementations, its classical applications, related open resources, followed by a discussion of open problems that are worth further research.

READ FULL TEXT

Robust Sequential Online Prediction with Dynamic Ensemble of Multiple Models: A Concise Introduction

Dynamic Ensemble Bayesian Filter for Robust Control of a Human Brain-machine Interface

META-DES.H: a dynamic ensemble selection technique using meta-learning and a dynamic weighting approach

Reinforcement Learning based dynamic weighing of Ensemble Models for Time Series Forecasting

Sequential online prediction in the presence of outliers and change points: an instant temporal structure learning approach

Classifier Pool Generation based on a Two-level Diversity Approach

RMSE-ELM: Recursive Model based Selective Ensemble of Extreme Learning Machines for Robustness Improvement

Ensemble distributional forecasting for insurance loss reserving

Robust Sequential Online Prediction with Dynamic Ensemble of Multiple Models: A Concise Introduction

Related Research

Dynamic Ensemble Bayesian Filter for Robust Control of a Human Brain-machine Interface

META-DES.H: a dynamic ensemble selection technique using meta-learning and a dynamic weighting approach

Reinforcement Learning based dynamic weighing of Ensemble Models for Time Series Forecasting

Sequential online prediction in the presence of outliers and change points: an instant temporal structure learning approach

Classifier Pool Generation based on a Two-level Diversity Approach

RMSE-ELM: Recursive Model based Selective Ensemble of Extreme Learning Machines for Robustness Improvement

Ensemble distributional forecasting for insurance loss reserving