Chapter 11
Parameter Control in Evolutionary
Optimisation
Margarita Antoniou, Rok Hribar, and Gregor Papa
Abstract Finding the global optimum of a complex function is one of the longstanding goals of applied mathematics and numerical analysis. Evolutionary algorithms have become a popular way of solving demanding and expensive optimisation problems. These algorithms are composed of several control parameters that
need to be set, for the procedure of searching for the optimum of an objective
function to be successful. Parameter setting is a challenging topic, since control
parameter values affect significantly the performance of the algorithm. Moreover,
the control parameters may interact with each other in an unpredictable way. On
top of this, at different stages of optimisation process, different control parameters
may be needed. In this chapter, we introduce some basic control parameters that
can be modified and the main methods of parameter settings, i.e. parameter tuning
(offline) and parameter control (online). More focus is given on parameter control
and the three different strategies used to implement it: the deterministic, the adaptive
and the self-adaptive parameter control. In addition, a short comparison between
parameter tuning and parameter control is given, based on the current literature. The
last section refers to the improvement that parameter control can bring in some of
the most complex instances of real-world optimisation problems, such as dynamic
or problems under uncertainty, when using EAs to solve them and some of the
strategies that can be used in each instance.
Keywords Parameter control · Parameter tuning · Evolutionary optimisation ·
Adaptation
M. Antoniou () · R. Hribar · G. Papa
Jožef Stefan Institute, Ljubljana, Slovenia
Jožef Stefan International Postgraduate School, Ljubljana, Slovenia
e-mail: margarita.antoniou@ijs.si; rok.hribar@ijs.si; gregor.papa@ijs.si
© Springer Nature Switzerland AG 2021
M. Vasile (ed.), Optimization Under Uncertainty with Applications to Aerospace
Engineering, https://doi.org/10.1007/978-3-030-60166-9_11
357
Parameter Control in Evolutionary
Optimisation
Margarita Antoniou, Rok Hribar, and Gregor Papa
Abstract Finding the global optimum of a complex function is one of the longstanding goals of applied mathematics and numerical analysis. Evolutionary algorithms have become a popular way of solving demanding and expensive optimisation problems. These algorithms are composed of several control parameters that
need to be set, for the procedure of searching for the optimum of an objective
function to be successful. Parameter setting is a challenging topic, since control
parameter values affect significantly the performance of the algorithm. Moreover,
the control parameters may interact with each other in an unpredictable way. On
top of this, at different stages of optimisation process, different control parameters
may be needed. In this chapter, we introduce some basic control parameters that
can be modified and the main methods of parameter settings, i.e. parameter tuning
(offline) and parameter control (online). More focus is given on parameter control
and the three different strategies used to implement it: the deterministic, the adaptive
and the self-adaptive parameter control. In addition, a short comparison between
parameter tuning and parameter control is given, based on the current literature. The
last section refers to the improvement that parameter control can bring in some of
the most complex instances of real-world optimisation problems, such as dynamic
or problems under uncertainty, when using EAs to solve them and some of the
strategies that can be used in each instance.
Keywords Parameter control · Parameter tuning · Evolutionary optimisation ·
Adaptation
M. Antoniou () · R. Hribar · G. Papa
Jožef Stefan Institute, Ljubljana, Slovenia
Jožef Stefan International Postgraduate School, Ljubljana, Slovenia
e-mail: margarita.antoniou@ijs.si; rok.hribar@ijs.si; gregor.papa@ijs.si
© Springer Nature Switzerland AG 2021
M. Vasile (ed.), Optimization Under Uncertainty with Applications to Aerospace
Engineering, https://doi.org/10.1007/978-3-030-60166-9_11
357