Full Program »
Markov Networks for Free ImprovisersThis paper discusses the use of probabilistic graphical models (PGMs) for initiating dynamical human musical interactions, in the context of free improvisation. I propose the more expressive model of Markov Networks and I speculate how they may serve for forming dynamical sepsets amongst players, based on their reciprocal beliefs expressed in the context of Bayesian inference. The prior is an assigned, private, musical personality. The players communicate their affinity preferences over a computer network using a graphical user interface implemented in the visual programming language Max, in order to represent the status of the undirected graph. The conclusion is that induced Markov Networks viewed as dynamical Bayesian games are employable in the context of free improvisation and distributed creativity, providing a valid (and conceptually very dissimilar) alternative to other structures that have been employed in music improvisation, such as graphic scores and idiom-based improvised forms.