ChanServ changed the topic of #mlpack to: "mlpack: a fast, flexible machine learning library :: We don't always respond instantly, but we will respond; please be patient :: Logs at http://www.mlpack.org/irc/
govg has quit [Ping timeout: 272 seconds]
govg has joined #mlpack
ImQ009 has joined #mlpack
betux22[m] has joined #mlpack
govg has quit [Read error: Connection reset by peer]
knightxerons has joined #mlpack
knightxerons has quit [Client Quit]
< shrit[m]1>
@rcurtin I am able now to pass most of the serialization test. The only blocking test is the hoeffding test.
< shrit[m]1>
There are two data members that are blocking me from serializing the hoeffding tree. First the dimensionMappings and the second is data::DatasetInfo
< shrit[m]1>
Both of them are std::unordered_map which has a pair inside
< shrit[m]1>
Cereal accept all C++ types (vector, map, tuples, etc...). However, I am afraid that it does not support nested types
< himanshu_pathak[>
Hey sakshamb189: are you there?
< sakshamb189[m]>
yes I am here
< himanshu_pathak[>
I have added test for mnist and circle dataset also. I am struggling with docs build.
< himanshu_pathak[>
Is there a way to build docs locally??
< sakshamb189[m]>
have you checked the console output
< sakshamb189[m]>
It fails while generating docs for kernel svm
< himanshu_pathak[>
Now I got where the problem is I am so bad at documentation
< himanshu_pathak[>
:)
< himanshu_pathak[>
It is failing because of C parameter
< sakshamb189[m]>
alright great now we can fix that ;)
< himanshu_pathak[>
yup I will fix it.
< sakshamb189[m]>
I see that the memory build is still failing?
< himanshu_pathak[>
Yeah problem with first test case problem with other test case has been solved. I am lookin into this
< himanshu_pathak[>
> Yeah problem with first test case problem with other test case has been solved. I am lookin into this
< himanshu_pathak[>
*already
< sakshamb189[m]>
alright sure. Let's try to fix that soon as well.
< himanshu_pathak[>
> alright sure. Let's try to fix that soon as well.
< himanshu_pathak[>
Maybe today if everything works fine also after this we have to compare it with sklearn also :)
< sakshamb189[m]>
Yeah let's do that.
< sakshamb189[m]>
Also once that is done do you think we can add support for multiple classes so, that we can run it on for example the entire mnist dataset?
< himanshu_pathak[>
> Yeah let's do that.
< himanshu_pathak[>
> Also once that is done do you think we can add support for multiple classes so, that we can run it on for example the entire mnist dataset?
< himanshu_pathak[>
Yup we can do that we have to train multiple binary svm for that as you have previously suggested
< sakshamb189[m]>
Yeah but currently in linear_svm they add parameters for all the svm models internally and train them all together. Is something like that possible with kernel svm?
< himanshu_pathak[>
> Yeah but currently in linear_svm they add parameters for all the svm models internally and train them all together. Is something like that possible with kernel svm?
< himanshu_pathak[>
I think that is when we are using it with linear svm with sgd in smo I think not but not sure I should look into this more first.
< sakshamb189[m]>
alright sure
< himanshu_pathak[>
anything else you want to discuss
< sakshamb189[m]>
If there are any other concerns you have as of now you can let me know.
< sakshamb189[m]>
otherwise we can meet next week
< himanshu_pathak[>
Docs build fixed thanks for the help I got stuck somwhere I will comment on my pr:)
< himanshu_pathak[>
*if I got stuck
< sakshamb189[m]>
alright great. Then we will meet next. Have a great week till then :)
< sakshamb189[m]>
next time*
< himanshu_pathak[>
Yup, have a great week ahead :)