6233RE: Starting with Neat, the XOR problem, some questions
- Jan 6, 2014
Reading and re-reading section 3.3 Protecting Innovation through Speciation, I want to make sure I have the correct concept because it appears to very computationally expensive. My understanding of the process is that you have to calculate the distance measure for each member of the population against every other member of the population to work out which members of the population are the same species? Eg something like below...
For each member of the population
For every other member of the population
Calculate the distance measure
This is very expensive because for each pair we need to go through and find the matching genes and calculate the weight differences of the matching genes
Once you have the distance measure of every member against every other member you can start to speciate. For all pairs that have a distance measure < x they go into the same species. So if for example your initial population is 100, you may have 50 members that go into the same species. For the remaining 50, you need to start the loop again because you know that these 50 do not fit into the first species but they may also not be the same species. So, for the remaining 50 you start the loop again and you find that maybe 30 of the 50 are in the same species, these 30 become the 2nd species. You now have 20 members left without a species so you start the loop again and continue until each member is assigned a species.
Is this correct? I will implement it and try it, but I can see if the population size is massive and the gene size is also massive it can start to take a long time to speciate each population member...
- << Previous post in topic Next post in topic >>