Brain-computer Interface for Assisting Decision-making on Individual Preference by Switching Support Vector Machines

Brain imaging techniques contribute to developments of neuro-economics and braincomputer interface (BCI). Although most of BCI systems were designed for assisting human motions, some BCIs were proposed as a system for assisting individual decision-making. However, due to lack of accuracy that the BCIs discriminate targets, their performance is still not enough for practical use. Hence, we propose a method for enhancing the accuracy by switching some learning machines automatically. As a device for obtaining brain information, we use NIRS, which is noninvasive and has potential for the miniaturization. And, we use support vector machines as the learning machines. By using the proposed BCI, experiments on preference were conducted and experimental results obtained from 7 participants have shown increase of 10% on the discrimination ratios of the proposed method comparing with a BCI with single leaning machine.