C4DM Seminar: Shlomo Dubnov - Information theoretic creativity, or how to find the optimal automata for music?
For external participants: Please join our mailing list to receive announcements of future C4DM seminars.
Date and Time Tuesday, 23th February 2016, at 11:00am
Place ITL Top Floor Meeting Room, Informatics Teaching Laboratory building, Queen Mary University of London, Mile End Road, London E1 4NS. Information on how to access the school can be found at here.
Speaker Shlomo Dubnov
Title Information theoretic creativity, or how to find the optimal automata for music?
Abstract In recent years there was a significant advance in using statistical and machine learning tools for creating musical models that are capable of generating music in a particular style. In the talk I will survey some of the recent works on controlling or "guiding" such models according to user specifications. Specifically I will describe the work done in UCSD / CREL on Variable Markov Oracle modeling that allows optimal model selection for audio signals, with applications to motif detection and guided improvisation, as well as other time series applications*. The general framework of our work is rooted in information theoretical approaches that use notions of complexity and accuracy to combine analytical and generative models in a common framework.
Bio Shlomo Dubnov is a Professor in UCSD Music Department and affiliate faculty in Computer Science and Engineering. He graduated the Rubin Music Academy in Jerusalem in Composition and holds a PhD in Computer Science from the Hebrew University, after which he served as an invited researcher in IRCAM, Paris. He directs UCSD Qualcomm's Institute Center for Research in Entertainment and Learning and is currently on a research excellence fellowship at LaBRI, Bordeaux.