| Abstract: | In recent years,machine learning techniques have been intensively used in functional magnetic resonance imaging (fMRI) data to decode visual stimuli, mental states, emotion and other information of interest. However, these studies are greatly hampered by the large number of predictor variables (voxels) and low observations (subjects). As a result, feature extraction techniques are used to remove redundant predictor variables and experimental noise for avoiding overfitting problem of machine learning model and enhancing model prediction accuracy and generalization ability. In this review, principles of some fMRI data supervised feature extraction techniques and current situation are briefly introduced, and we focus on discussing their performance and possible improvement of methods. In the end, the future research directions are prospected. |
