Optimisation using population based incremental learning (PBIL)

The population-based incremental learning (PBIL) algorithm is a simple stochastic optimisation technique that can be applied quickly to a wide range of problems. The technique's main area of application is for problems that are too multimodal or discontinuous for gradient or simplex methods, but don't warrant a full evolutionary algorithm solution. PBIL has been shown to outperform conventional deterministic and stochastic optimisation techniques on a wide range of problems and yet is simple to code. This paper describes a practical approach to applying the PBIL algorithm to optimisation problems. First the operation of the algorithm is described and then guidelines for tuning the algorithm are presented. An example implementation is given and a method for reducing the processing burden of the algorithm is detailed. An example MATLAB routine is included to demonstrate the simplicity of the algorithm.