Skip to main content

A neural networks based model for automatic performance of musical scores


The aim of the project is the development of a general concept for local search operators. The identification of optimal operators for solving a given combinatorial optimisation problem is fundamental for an efficient local search approach. This is true for both, point-based local search methods (local hillclimbing, simulated annealing) as well as for population-based local search methods (genetic algorithms). in the first place, the concept deals with unary operators as used in point-based local search methods. However, since recombination operators as usually found in genetic algorithms, can be traced back to unary operators the concept also applies to genetic algorithms and thus provides a general framework. In particular, such a concept is expected to improve the current design methodology for local search algorithms. Furthermore, it may give rise to an alternative theoretical explanation for genetic algorithms.


Kungliga Tekniska Högskolan

100 44 Stockholm

Participants (1)

Not available