Consider a genetic algorithm rule extraction algorithm that uses the Pittsburg representation. Which of the following approaches helps to reduce overfitting, while still ensuring that the individuals will converge to one solution $($i$.$e$.$ one rule set$).$ Incorrect answers will be penalized.

Question $6$Select one or more:

a$.$

A new mutation operator that will delete a randomly selected rule from an individual

b$.$

This algorithm does not suffer from overfitting, so none of the above

c$.$

A high probability of mutation, where mutation simply flip bits

d$.$

A crossover operator where the Boolean AND operator is applied on the bits of the two parents between two randomly selected cross$-$over points

e$.$

A crossover operator where the Boolean OR operator is applied on the bits of the two parents between two randomly selected cross$-$over points

f$.$

Add a penalty function to the fitness function, where the penalty is simply the sum over all of the bit values of the chromosome