On the Optimality of the Simple Bayesian Classifier under Zero-One Loss
Machine Learning - Special issue on learning with probabilistic representations
Hi-index | 0.00 |
This paper proposed a new method, combining EEG and psychology instruments, to detect stress which can contribute in prediction and intervention of major depression. Seven mothers with mental retardation children as stress group and four age-matched mothers with healthy children as normal controls are enlisted. Results showed that relative power in alpha rhythm of stress group is significantly less than normal controls, while relative power in theta rhythm is much larger than normal controls. Discrimination accuracy gets higher than only using psychology instruments for distinguishing the two groups in our experiment. Besides, combination of EEG linear and nonlinear features is better than using only linear ones. Combination of LZ-complexity, alpha relative power and PSQI achieves discrimination accuracy of 95.12%, which gains an improvement of 19.51% compared with accuracy by using only PSQI. As a result, the combination of EEG and psychology instruments will benefit the detection of stress.