Bayesian approaches to failure prediction for disk drives

Bayesian approaches to failure prediction for disk drives (2001)

by Greg Hamerly , Charles Elkan

Venue:	In Proc. 18th ICML
Citations:	20 - 2 self

Datum	Value	Source
TITLE	Bayesian approaches to failure prediction for disk drives	SVM HeaderParse 0.2
AUTHOR NAME	Greg Hamerly	SVM HeaderParse 0.2
AUTHOR NAME	Charles Elkan	SVM HeaderParse 0.2
AUTHOR AFFIL	Department of Computer Science and Engineering, University of California, San Diego	SVM HeaderParse 0.2
AUTHOR ADDR	La Jolla, California 92093-0114	SVM HeaderParse 0.2
ABSTRACT	Hard disk drive failures are rare but are often costly. The ability to predict failures is important to consumers, drive manufacturers, and computer system manufacturers alike. In this paper we investigate the abilities of two Bayesian methods to predict disk drive failures based on measurements of drive internal conditions. We first view the problem from an anomaly detection stance. We introduce a mixture model of naive Bayes submodels (i.e. clusters) that is trained using expectation-maximization. The second method is a naive Bayes classifier, a supervised learning approach. Both methods are tested on realworld data concerning 1936 drives. The predictive accuracy of both algorithms is far higher than the accuracy of thresholding methods used in the disk drive industry today. 1.	SVM HeaderParse 0.2
YEAR	2001	INFERENCE
VENUE	In Proc. 18th ICML	INFERENCE
VENUE TYPE	CONFERENCE	INFERENCE
CITATIONS	14 found	ParsCit 1.0