### Large Scale Image and Music Annotation: Learning to Rank and Multi-Tasking with Joint Embeddings

par/by Jason Weston
In the first part of the talk we will discuss large scale image annotation. Image annotation datasets are becoming larger and larger, with tens of millions of images and tens of thousands of possible annotations. We propose a well performing method that scales to such datasets by simultaneously learning to optimize precision at k of the ranked list of annotations for a given image {\em and} learning a low-dimensional joint embedding space for both images and annotations. Our method both outperforms several baseline methods and, in comparison to them, is faster and consumes less memory. We also demonstrate how our method learns an interpretable model, where annotations with alternate spellings or even languages are close in the embedding space. Hence, even when our model does not predict the exact annotation given by a human labeler, it often predicts similar annotations, a fact that we try to quantify by measuring the newly introduced sibling'' precision metric, where our method also obtains good results. In the second (shorter) part of the talk we will discuss large scale music annotation. Music prediction tasks range from predicting the genre, style or the artist given a song or clip of audio, predicting similar artists given an artist, or predicting related songs given a song, clip, artist name or genre or style tag. That is, we are in interested in pretty much every semantic relationship between the different musical concepts in our database. In realistic databases, the number of songs is measured in the millions, and the number of artists in the tens of thousands or more, providing a considerable challenge to standard machine learning techniques. In this work, we propose a method that scales to such datasets which attempts to capture the semantic similarities between the database items by modeling audio, artist names, and genre and style tags in a single low-dimensional semantic space. This choice of space is learnt by optimizing the set of predictions tasks of interest jointly using multi-task learning. Our method both outperforms baseline methods and, in comparison to them, is faster and consumes less memory. We then demonstrate how our method learns an interpretable model, where the semantic space captures well the similarities of interest.