verne.freenode.net changed the topic of #mlpack to: http://www.mlpack.org/ -- We don't respond instantly... but we will respond. Give it a few minutes. Or hours. -- Channel logs: http://www.mlpack.org/irc/
syrius has quit [Remote host closed the connection]
syrius has joined #mlpack
KTL has quit [Ping timeout: 245 seconds]
dataVinci has joined #mlpack
dataVinci has quit [Ping timeout: 250 seconds]
< zoq>
"Submissions will be reviewed based on the status of the project at the time of the submission deadline.", I guess the criterion is fixed
< naywhayare>
yeah, but I don't think that means we have to rush a 1.1.0 release; they'll probably look at github as of April 28th
< zoq>
hm, okay
< zoq>
Thanks!
< naywhayare>
yeah, my plan is to reference the abstract to the git master branch (and mention the features will be released in 1.1.0 when it comes out)
< zoq>
sounds like a good plan
< zoq>
JaraxussTong: Hey, I've made some changes to the convolution methods and pushed the changes to the master branch. It seems to be about 3.5 times faster to use pointers instead of using kernel(ki,kj). As you already pointed out the FFT method isn't really fast for small kernels, I improved the computation slightly, but the bottleneck is armadillo's fft implementation. There is also a new convolution method whic
< zoq>
h uses svd to speed up the computation.
< zoq>
JaraxussTong: Ideally I would like to proceed with small chunks instead of merging everything at once. Since some things changed, I'll go and make the necessary changes, so you don't need to worry about that. Maybe you can create a new pull request that contains the pooling rules only? If not I'll go and cherry pick the files. Ideally I like to avoid to copy the files and commit the files under my name because
< zoq>
it's your work and you deserve the credits.