Background process' fitness function
where:
T
=
training set
pt(a,a')
= {
0.0 if
a
loses against
a'
0.5 if
a
and
a'
tie
1.0 if
a
wins against
a'
l(a')
=
number of games lost by
a'
