Skip to main content

A-Fitness-Function-for-Creativity-in-Jazz-Improvisation-and-Beyond (Jordanous 2010) - code

Jordanous, Anna (2010) A-Fitness-Function-for-Creativity-in-Jazz-Improvisation-and-Beyond (Jordanous 2010) - code. github Java code. (KAR id:42882)


Can a computer evolve creative entities based on how creative they are? Taking the domain of jazz improvisation, this work investigates how creativity can be evolved and evaluated by a computational system. The aim is for the system to work with minimal human assistance, as autonomously as possible. The system employs a genetic algorithm to evolve musical parameters for algorithmic jazz music improvisation. For each set of parameters, several improvisations are generated. The fi?tness function of the genetic algorithm implements a set of criteria for creativity proposed by Graeme Ritchie. The evolution of the improvisation parameters is directed by the creativity demonstrated in the generated improvisations. From preliminary fi?ndings, whilst Ritchie's criteria does guide the system towards producing more acceptably pleasing and typical jazz music, the criteria (in their current form) rely too heavily on human intervention to be practically useful for computational evaluation of creativity. In pursuing more autonomous creativity assessment, however, this system is a promising testbed for examining alternative theories about how creativity could be evaluated computationally.

Item Type: Software
Subjects: M Music and Books on Music > M Music
Q Science > Q Science (General) > Q335 Artificial intelligence
Q Science > QA Mathematics (inc Computing science) > QA 76 Software, computer programming, > QA76.87 Neural computers, neural networks
Divisions: Faculties > Sciences > School of Computing
Faculties > Sciences > School of Computing > Computational Intelligence Group
Depositing User: Anna Jordanous
Date Deposited: 10 Sep 2014 11:21 UTC
Last Modified: 29 May 2019 13:04 UTC
Resource URI: (The current URI for this page, for reference purposes)
Jordanous, Anna:
  • Depositors only (login required):


Downloads per month over past year