Reputation
12,388
Top tag
Next privilege 15,000 Rep.
Protect questions
Badges
6 62 110
Newest
 Nice Answer
Impact
~936k people reached

Feb
4
awarded  Nice Answer
Jan
19
awarded  Favorite Question
Jan
18
awarded  Yearling
Jan
13
comment Help: Random Forest optimization (image classification)
@also try different numbers of candidate features at each split.
Jan
12
comment Help: Random Forest optimization (image classification)
Fit your model on a training set, and then calculate error on a test set that the model did not see during training: en.wikipedia.org/wiki/Test_set
Jan
12
comment Help: Random Forest optimization (image classification)
Try more trees (e.g. 4,000), deeper tress, and tuning the mtry parameter (or number of variables considered per split). Test all these parameters out of sample.
Jan
12
comment Help: Random Forest optimization (image classification)
Perhaps try a covnet instead of a random forest: github.com/fchollet/keras/blob/master/examples/cifar10_cnn.py
Dec
28
revised R libraries for deep learning
added 381 characters in body
Dec
26
awarded  Notable Question
Dec
16
awarded  Benefactor
Dec
16
comment What is “Multinomial Deviance” in the glmnet package?
Thank you so much. I have a much better understanding of this topic now.
Dec
16
comment What is “Multinomial Deviance” in the glmnet package?
Thanks for such an excellent answer! When you say "which is just the empirical log loss multiplied by a constant", what is the constant? Is it always the same, or does it vary problem by problem? Mentally I'm trying to figure out an easy way to convert the "multinomial deviance" scale to "multiclass logloss", which I have a much better intuitive understanding about.
Dec
16
comment What is “Multinomial Deviance” in the glmnet package?
Thank you for the excellent, detailed answer. One last question— how does this deviance function (which I think glmnet computes as "predictive" deviance on out-of-sample data) relate to "multi-class" logloss?
Dec
16
accepted What is “Multinomial Deviance” in the glmnet package?
Dec
14
revised What is “Multinomial Deviance” in the glmnet package?
deleted 17 characters in body
Dec
13
revised What is “Multinomial Deviance” in the glmnet package?
added 4 characters in body
Dec
13
asked What is “Multinomial Deviance” in the glmnet package?
Dec
8
comment What happens if the explanatory and response variables are sorted independently before regression?
People like to deceive themselves, and often become irritable when that deception is noted. A really important skill to learn in your career is how to gently counter that deception (try channeling the best elementary school teacher you every knew). Another important skill is identify when that deception is unshakable and avoiding those situations...
Dec
1
awarded  Necromancer
Nov
24
revised Vowpal Wabbit: best strategy for short text data like titles & kewords
deleted 1 character in body