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/
kris1 has left #mlpack []
kris1 has joined #mlpack
chenzhe has quit [Ping timeout: 246 seconds]
kris1 has quit [Quit: kris1]
chenzhe has joined #mlpack
chenzhe has quit [Ping timeout: 255 seconds]
sgupta has quit [Ping timeout: 240 seconds]
sgupta has joined #mlpack
sumedhghaisas has quit [Ping timeout: 260 seconds]
vivekp has joined #mlpack
vivekp has quit [Ping timeout: 255 seconds]
vivekp has joined #mlpack
shikhar has joined #mlpack
vivekp has quit [Ping timeout: 260 seconds]
vivekp has joined #mlpack
kris1_ has joined #mlpack
govg_ has quit [Ping timeout: 240 seconds]
govg has joined #mlpack
sgupta has quit [Ping timeout: 260 seconds]
shikhar has quit [Quit: WeeChat 1.7]
sgupta has joined #mlpack
shikhar has joined #mlpack
kris1_ has quit [Quit: kris1_]
kris1_ has joined #mlpack
< zoq>
kris1_: Sounds like Mikhail fixed the problems :)
< kris1_>
Yes i was talking to him…..he said he would comment in an hour
< kris1_>
Curious to know what was going wrong
< kris1_>
Ohh i just got the mail
< kris1_>
I will have a look now
< zoq>
Haven't looked at the patch yet, but setting cumgradient to zero sounds like a issue. Also really hard to find since you don't end up with an error.
< zoq>
So, nice catch!
kris1_ has quit [Quit: kris1_]
< lozhnikov>
zoq: That was not the main issue, the sign of the gradient was invalid.
< zoq>
lozhnikov: ah, easy to miss, for me it's often helpful to sit down and do the calculation manually.
aashay has quit [Quit: Connection closed for inactivity]
< zoq>
rcurtin: The mail setup for the models repo should be ready right? Not sure I just missed the last mail somehow.
chenzhe1 has joined #mlpack
chenzhe has quit [Ping timeout: 246 seconds]
chenzhe1 is now known as chenzhe
mikeling has quit [Quit: Connection closed for inactivity]
aashay has joined #mlpack
kris1 has quit [Ping timeout: 240 seconds]
kris1 has joined #mlpack
< rcurtin>
zoq: hmm, let me check when I get home
kris1 has quit [Quit: kris1]
kris1 has joined #mlpack
kris1 has quit [Quit: kris1]
kris1 has joined #mlpack
chenzhe has quit [Ping timeout: 246 seconds]
chenzhe has joined #mlpack
chenzhe has quit [Ping timeout: 240 seconds]
< kris1>
How can i get the full mnist dataset for reading with armadillo
< kris1>
Is there a full mnist dataset.arm available
< lozhnikov>
If I am not mistaken armadillo supports csv
< kris1>
hmmm yes it does but i think the arm would be comprssed or something
< kris1>
i will download the csv and try them
< kris1>
then
sumedhghaisas has joined #mlpack
chenzhe has joined #mlpack
mentekid has quit [Quit: Leaving.]
chenzhe has quit [Ping timeout: 246 seconds]
< kris1>
So i tested out rbm implementation with thyroid dataset
< kris1>
i tested classification acurracy with ffn
< kris1>
i am getting aroung 93%
< kris1>
and with ffn we are getting aroung 94%
< kris1>
the dimension of the dataset with rbm representation is around 10.
< zoq>
I would recommend to test another dataset the thyroid is unbalanced, so 93/94% isn't that good. Do you have results for the mnist dataset?
sgupta has quit [Ping timeout: 240 seconds]
< kris1>
Well i am sorry the represtation was still 21 but the accuraccy was 93 %
< kris1>
i am write now working to get the accuray for the 18 case
< kris1>
for mnist we got really bad samples the reason for this is that there are only 250 data points which are very less for any deep learning system
< kris1>
i tried with whole mnist dataset but it would takes a lot of time. deeplearning.net reported 122.466 minutes on a Intel Xeon E5430 @ 2.66GHz CPU
< zoq>
I guess, you are talking about the mnist subset, since the complete dataset has about 60000 training samples.
< kris1>
yes the dataset provied in the mlpack data folder
< zoq>
you could create a new subset from the complete mnist dataset
< kris1>
Well we don’t exactly now how many examples we would need it has to be hit and trial.
< kris1>
But aroung 20k would give a reasonable result i think