Fault Diagnosis Analysis in Large-Scale Computing Environments

This paper issues the problem of fault diagnosis in high computing system. In order to solve this problem, i.e., correctly and efficiently detecting the anomaly nodes during the system operation, which is very similar to the principle of pattern recognition research work, thus we try to use some pattern recognition methods to analysis and solve fault diagnosis problem in this paper. And also we do some experiment and compare the results and finally get some useful conclusion to show that Kernel Eigenface and Kernel Fisherface methods achieve lower error rates than the ICA and PCA approaches in anomaly nodes detection.