Search code examples
artificial-intelligenceneural-networkgenetic-algorithm

In a genetic algorithm is roulette wheel selection the same as Monte Carlo method?


After reading a paper they mention the Monte Carlo method for selecting fittest chromosomes. Is this the same as roulette wheel selection?

Direct quote

The same initial population of individuals was used for each problem and architecture. The population sizes were 150 individuals (except for HD problem that was 100). The following techniques were employed: the Montecarlo method for the selection of individuals; the Darwinian substitution method; a single crossover point; a crossover rate of 90%; and a mutation rate of 10%.


Solution

  • I'm reading this paper by Goldberg & Deb (1991). In it they have also mentioned the term Monte Carlo, exact sentence is (Page 3: Section "Proportionate Reproduction"):

    ... Various methods have been suggested for sampling this probability distribution, including Monte Carlo or roulette wheel selection (De Jong, 1975), stochastic remainder selection (Booker, 1982; Brindle, 1981).....

    So I suspect the term Monte Carlo was used to denote they used the method to generate a random number as a part of the selection process, rather than using it solely for selection.

    If you remember computers back in the day needed a good random number generator, so they could have been using Monte Carlo for that.