News

Representing Style: the contribution of Markov Constraints

Friday 27 April 2012 - Francois Pachet

ARGOMENTI: Seminari

SEMINARI DI INFORMATICA
Friday 27 April 2012 h. 11:30, room 2BC30
Francois Pachet (SONY Computer Science Laboratory, Paris)
"Representing Style: the contribution of Markov Constraints"

-Abstract
The motivation of this work is to generate sequences (typically music or text) that mitate a given -style-. A traditional approach to represent style is to use tatistical methods such as Markov processes. Markov processes capture dequately short-term dependencies in temporal sequences, but are very difficult to control. Imposing arbitrar properties on Markov chains is indeed paradoxical as it often goes in the way of the basic Markov hypothesis of limited temporal dependency. This talk introduces our work on Markov Constraints: a new class of global constraints whose goal is to reformulate Markov processes in the framework of CSP, to enable users to generate sequences -in the style of- a given corpus, while satisfying arbitrary control criteria. We will give examples in music and text generation, and relate this work to the ERC funded Flow Machines projects.

-References:
http://www.csl.sony.fr/~pachet/flow_machines.html
Pachet, F. and Roy, P. Markov constraints: steerable generation of Markov equences. Constraints, 16(2):148-172 March 2011.
Pachet, F., Roy, P. and Barbieri, G. Finite-Length Markov Processes with Constraints. Proceedings of the 22nd International Joint Conference on Artificial Intelligence, IJCAI, pages 635-642, Barcelona, Spain, July 2011

-Bio
François Pachet received his Ph.D. and Habilitation degrees from Paris 6 University (UPMC). He is a Civil Engineer (Ecole des Ponts and Chaussées) and was Assistant Professor in Arti?cial Intelligence and Computer Science, at Paris 6 University, until 1997. He then set up the music research team at SONY Computer Science Laboratory Paris, where he developed the vision that metadata can greatly enhance the musical experience in all its dimensions, from listening to performance. His team conducts research in interactive music listening and performance and musical metadata and developed several innovative technologies (constraint-based spatialization, intelligent music scheduling using metadata) and award winning systems (MusicSpace, PathBuilder, The Continuator for Interactive Music Improvisation, etc.). His current research focuses on creativity and content generation, as he was recently awarded an ERC Advanced Grant to develop the concepts and technologies of "flow machines": a new generation of content generation tools that help users find and develop their own "style".

Download Seminari di Informatica

NEWS: New Second Level Degree in Data Science - Second cycle degree - a. y. 2017/18 X