抄録
In recent years, with the development of IoT technologies, biological information of users has been easily acquired and accumulated. It is required that user's cognitive or emotional state could be accurately identified from such biological information. Various researches have used machine learning or deep learning however, generalized methods to handle biological data have not been established. In this paper, we propose a more versatile method using convolution neural network of deep learning to extract features from individual biometric information which is multi-channels time-series data representing user’s cognitive. We demonstrate a new method of extraction and distinguishment of features from cerebral blood flow, skin temperature, respiration and skin conductance measured during specific task conditions in the stroop experiment.
