Ranking method inconsistency between duplicated projects

[max-markov]

Hello,

In the WANN source, it seems you are ranking the individuals as described in the paper: by mean fitness + num connections (80%), or by peak fitness + num connections (20%).

In the prettyNeatWann source, however, you are sorting based only on mean fitness (80%) or by mean fitness + num connections (20%).

Is there a reason why this is different between the sources? And in general, what is the motivation of the separation between the WANN and prettyNeatWann sources? Couldn't fully understand this from the Readme description.

Thank you!

[agaier]

Hi, There are different ways or parameterizing the algorithm I was testing -- there is no reason the sorting is different, just an oversight on my part in not unifying them. The motivation of having two sources was that the WANN code would be more self contained and easier to understand (so aimed at people who read the paper and wanted to more details about implementation), and the prettyNeatWann code would be easier to extend and do experiments with (so aimed at people who want to actually use the code).

…

On Thu, Jan 23, 2020, 12:51 max-markov, ***@***.***> wrote: Hello, In the WANN source, it seems you are ranking the individuals as described in the paper: by mean fitness + num connections (80%), or by peak fitness + num connections (20%). In the prettyNeatWann source, however, you are sorting based only on mean fitness (80%) or by mean fitness + num connections (20%). Is there a reason why this is different between the sources? And in general, what is the motivation of the separation between the WANN and prettyNeatWann sources? Couldn't fully understand this from the Readme description. Thank you! — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#14?email_source=notifications&email_token=AAVWKJ7VPHCXYYXFYLMKGHDQ7GAFHA5CNFSM4KKVFUS2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4IIHBSMA>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAVWKJ67SL3Y23N4E33JOM3Q7GAFHANCNFSM4KKVFUSQ> .