Decoding color of stimuli given to a human subject from functional magnetic resonance imaging voxel patterns using machine learning algorithm

A brain decoding of visual stimuli using various machine learning is proposed in order to make a foundation of brain computer interface. Visual stimuli that are representations of objects, shapes, colors, and so on, are important information for human perception. Some of properties of processing of visual information in human brain are revealed, for example existence of neuron responding an orientation of line segment. This research reveals the precision of pattern recognition using supervised machine learning of human brain activity when human see color circle drawn In a display. Support vector machine with various kernel, neural network, random forest, and sparse logistic regression are employed in this research and compared among each other. The result shows that the highest precision Is 71% for predicting color of circle from three colors using sparse logistic regression.