In other words, the research from the University of Jyväskylä, indicates that the way you dance is unique, and from the subtle differences between dance patterns, algorithms can tell it’s you rather than someone else.
The objective of the research was to apply machine learning to understand how and why music affects people the way that it does.
To explore this question, the Finnish scientists used motion capture technology (much like the technology now common movies with a CGI element) to gain an insight about the uniqueness of dance moves and to also extrapolate what the dance move might say about the person.
From studying different patterns of dancing, the researchers are of the view that they can determine how extroverted or neurotic a person is and also draw insights in the particular mood a person is experiencing.
The recent study used seventy-three people, who were motion captured dancing to eight different forms of music: Blues, Country, Dance/Electronica, Jazz, Metal, Pop, Reggae and Rap.
The captured digital data was passed through a machine learning algorithm. This revealed that the computer could correctly identify which of the 73 individuals was dancing 94 percent of the time, irrespective of the music genre that the person was dancing to.
This led one of the researchers, Dr. Pasi Saari, to summarize: “It seems as though a person’s dance movements are a kind of fingerprint… Each person has a unique movement signature that stays the same no matter what kind of music is playing.”
There were some differences in terms of accuracy in relation to music genera. The algorithm had a harder time, for example, identifying people dancing to metal music. This might be, the researchers speculate, because most people default to a standard ‘head-banging’ style when dancing to metal.
For the next stage of the research, the scientists want to see whether musical style changes over a person’s lifetime, and also whether there are significant cultural differences. The study will also see if humans can recognize another person from their dancing, or if the algorithm is in advance of human capabilities.
The study has been published in the Journal of New Music Research. The research paper is titled “Dance to your own drum: Identification of musical genre and individual dancer from motion capture using machine learning.”
