The article Experimenting with Musically Motivated Convolutional Neural Networks has obtained the Best Paper Award at CBMI2016, the 14th International Workshop on Content-Based Multimedia Indexing. The authors explore various architectural choices of relevance for music signals classification tasks in order to start understanding what the chosen deep learning based algorithms can learn from a particular set of data. 

Jordi Pons, first author, conducts his PhD in the context of the subproject Machine learning approaches for structuring large sound and music collections, led by Xavier Serra. This link provides access to code and dataset used, in addition to article.