object ve u ct o
cu ve s co p e
w t
o e t a
o e sadd e
e problem becomes complex. This is because it is impossible to
e that all the saddle points have the same minimum objective
value. In addition, the Newton’s method can only guarantee an
ion process to reach one saddle point based on one initial point
ective function curve. Therefore, one initial point leads to one of
saddle points using the Newton’s method. Whenever one saddle
been reached, the Newton’s optimisation process will never be
move out from the saddle point even if this saddle point does not
nd to the minimum objective function value.
fore, the evolutionary computation techniques or algorithms have
sidered for the optimisation problems where a problem requires
x objective function curve with multiple saddle points [Fogel, et
; Holland, 1975; Goldberg, 1989]. The basic principle of the
ary computation approaches is to start from multiple initial
an objective function curve. Multiple models are constructed
e Newton’s method or similar greatest descent methods
eously from these multiple initial points. In an evolutionary
ion model, a pool of candidates (or candidate solutions) is
Within these candidates, the principle of survival of the fittest is
Only those candidates with the best fitness measurements have
to maintain in the pool and have the right to breed new offspring,
e used to replace those which fail the designed fitness criterion.
this kind of evolution process, an optimal candidate or solution
em can be discovered finally. The evolutionary computation
es have been widely used in many biological/medical research
with success [Fogel, 2008; Francois and Siggia, 2008; Rostain, et
An, et al., 2020; Eremeev and Spirov, 2021].
chapters
k is composed of nine chapters. From Chapter 2 to Chapter 8,
apter is designed to introduce a specific biological pattern