Skip to main content

Table 2 Parameter settings and brief descriptions

From: Evolution of control with learning classifier systems

Parameter

Value

Brief Description

R

790

Rule population size

γ

0.76

Discount rate

θ mna

6

Min. number of actions in match set

P #

0.4

Probability of hash

p I

7.4

Initial payoff

ε I

1.0

Initial error

F I

0.03

Initial fitness

ε 0

18.5

Error threshold

θ ga

260

Genetic algorithm frequency

θ del

32

Deletion threshold

β

0.01

Affects update of p, ε, and action set size for classifiers

α

0.087

Affects fitness updates

ν

0.01

Affects fitness updates

χ

0.711

Likelihood of GA crossover operation

μ

0.263

Likelihood of GA mutation operation

δ

0.05

Modifies the effect of fitness on the ‘deletion vote’ of a classifier

θ sub

31.579

Subsumption threshold

p explr

N/A

Likelihood of exploring

doActionSetSubsumption

true

Perform subsumption in the action set?

doGaSubsumption

true

Perform subsumption in the GA?

useNewActionChooser

true

Use the new action chooser mechanism?