ABOUT

This talk covers several technical approaches that brought a music recommendation service into the AI era. In particular, this is a case study of a collaboration between machine learning experts (Unbox Research) and a music service company (Roon Labs). By implementing new personalization features, our joint work was able to significantly improve product quality and user engagement. This talk begins with the intuition behind the collaborative filtering (CF) method, which assigns a numeric representation to each user and product (this idea applies to basically any kind of content — not just music). The speaker will also touch upon several more advanced techniques that can augment a traditional CF personalization pipeline.