Learning-aided predictor integration for system performance prediction

doi:10.1007/s10586-007-0041-8

Learning-aided predictor integration for system performance prediction

2007-10-02

5

PRE

AI

DOI：10.1007/s10586-007-0041-8

Original

Original request for help

Save

Abstract

En 中文

The integration of multiple predictors promises higher prediction accuracy than the accuracy that can be obtained with a single predictor. The challenge is how to select the best predictor at any given moment. Traditionally, multiple predictors are run in parallel and the one that generates the best result is selected for prediction. In this paper, we propose a novel approach for predictor integration based on the learning of historical predictions. Compared with the traditional approach, it does not require running all the predictors simultaneously. Instead, it uses classification algorithms such as k-Nearest Neighbor (k-NN) and Bayesian classification and dimension reduction technique such as Principal Component Analysis (PCA) to forecast the best predictor for the workload under study based on the learning of historical predictions. Then only the forecasted best predictor is run for prediction. Our experimental results show that it achieved 20.18% higher best predictor forecasting accuracy than the cumulative MSE based predictor selection approach used in the popular Network Weather Service system. In addition, it outperformed the observed most accurate single predictor in the pool for 44.23% of the performance traces.

Keywords:

system performance

virtual machine

virtual machine monitor

k-nearest neighbor (kNN)

Bayesian classification

principal component analysis (PCA)

time-series prediction

Learning-aided predictor integration for system performance prediction

Abstract

AI Summary

Journal

Organization