Booth Id:
ROBO046
Category:
Robotics and Intelligent Machines
Year:
2019
Finalist Names:
Ochsner, Tobia (School: Kantonsschule Schaffhausen)
Abstract:
It has never been easier to listen to one's favourite music than today: With services like Spotify or Apple Music, we have access to an almost unlimited number of songs. But this variety can be overwhelming sometimes. The goal of my project was the creation of an artificial intelligence which is able to generate playlists based on a given song. I used the so-called neural networks which are widely used in domains such as object or speech recognition. 2000 playlists of Spotify (for example "Pop Rising" or "Summer Acoustic") were used as examples to train the neural networks. I developed novel methods for Training (including a novel loss function and a unique combination of several networks) and was able to achieve a significant increase in performance compared to other methods.