Machine learning-based classification analysis of knowledge worker mental stress.

The aim of this study is to analyze the performance of classifying stress and non-stress by measuring biosignal data using a wearable watch without interfering with work activities at work. An experiment is designed where participants wear a Galaxy Watch3 to measure HR and photoplethysmography data while performing stress-inducing and relaxation tasks. The classification model was constructed using k-NN, SVM, DT, LR, RF, and MLP classifiers. The performance of each classifier was evaluated using LOSO-CV as a verification method. When the top 9 features, including the average and minimum value of HR, average of NNI, SDNN, vLF, HF, LF, LF/HF ratio, and total power, were used in the classification model, it showed the best performance with an accuracy of 0.817 and an F1 score of 0.801. This study also finds that it is necessary to measure physiological data for more than 2 or 3 min to accurately distinguish stress states.

An optimization criterion for generalized discriminant analysis on undersampled problems.

An optimization criterion is presented for discriminant analysis. The criterion extends the optimization criteria of the classical Linear Discriminant Analysis (LDA) through the use of the pseudoinverse when the scatter matrices are singular. It is applicable regardless of the relative sizes of the data dimension and sample size, overcoming a limitation of classical LDA. The optimization problem can be solved analytically by applying the Generalized Singular Value Decomposition (GSVD) technique. The pseudoinverse has been suggested and used for undersampled problems in the past, where the data dimension exceeds the number of data points. The criterion proposed in this paper provides a theoretical justification for this procedure. An approximation algorithm for the GSVD-based approach is also presented. It reduces the computational complexity by finding subclusters of each cluster and uses their centroids to capture the structure of each cluster. This reduced problem yields much smaller matrices to which the GSVD can be applied efficiently. Experiments on text data, with up to 7,000 dimensions, show that the approximation algorithm produces results that are close to those produced by the exact algorithm.