Music streaming sites base song recommendations on users’ playback history. The process employed is called collaborative filtering. With this method, recommendations are based on the data of other users with similar playback histories, so sometimes users get song recommendations that do not suit their tastes in music. I would like to improve this process so that users get recommendations that are more to their individual preferences.
　For this research, I am making use of users’ brain waves. When we listen to music, we make unconscious judgments based not only on genre, but also a variety of other factors, such as rhythm, voice and tone. How such factors affect preferences vary on the individual responses of listeners. I thought it would be useful to use encephalography to learn how these factors are related to listener responses.
　Ordinarily, brain waves are measured as a time series. However, it is easier to characterize responses when the series is converted to frequency (Hz) domains. (It is thought that the percentage of high frequency waves is higher when the brain is in an aroused state, and that the percentage of low frequency waves is higher when the brain is relaxed.） By accumulating data converted to frequency domains and making use of machine learning, it becomes possible to recommend music that elicits emotional responses. That’s my study interest in a nutshell.

　I really like music. A person can only listen to a certain number of pieces in one lifetime! Speaking hyperbolically, a person might die before they encounter what to them is the “best song.” If this research helps identify the songs that best suit the tastes of each individual, we all will have an opportunity to find our “best songs.” I just think that would be marvelous.
　Department of Computer Science and Communication Engineering encompasses not only my research, but also media content, security, network construction and a wide range of other fields. Anyone interested in the field of information and communication should be able to find a research topic that suits their interest.