[mlpack] Neural Evolution Algorithm - Week 5
bang liu
lbldtb at gmail.com
Tue Jun 28 01:51:32 EDT 2016
Dear all,
During last week, the NEAT algorithm is almost finished (remaining is
mainly decide how to organize all the mutate and crossover operations to
generate next generation of genome population. It is kind of flexible.)
During implementing NEAT, we implemented Population class which contains
multiple species of genomes; also, multiple previous classes are revised
for recording more information.
In this week, we are going to revise the coding of NEAT, as currently I
found that there are lots of thing to consider when class is big and
contains lots of functions. I am trying to keep class as simple as
possible, thus only add critical functions and members; besides, we also
want the running efficiency is high. Actually I found that it is kind of a
tradeoff.
Besides, we are also going to test NEAT by testing cases such as XOR and
pole balancing problem. Later it will be excited if it can run the Mario
game.
Best,
Bang
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.cc.gatech.edu/pipermail/mlpack/attachments/20160627/c3630be9/attachment.html>
More information about the mlpack
mailing list