Take the 2-minute tour ×
Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. It's 100% free, no registration required.

I have a dataset with 60,000 training examples. I tried the following command under windows and with libsvm:

svm-train.exe -t 2 -g 0.07 -c 1 images.train

and it started giving multiple outputs for example:

.....*..*
optimization finished, #iter = 7090
nu = 0.117216
obj = -667.069565, rho = -0.055205
nSV = 4702, nBSV = 26
....*..*
optimization finished, #iter = 6485
nu = 0.100224
obj = -565.615150, rho = -0.198917
nSV = 4342, nBSV = 29
..*.*
optimization finished, #iter = 3671
nu = 0.027754
obj = -176.698407, rho = -0.838783
nSV = 2560, nBSV = 52
....*..*
optimization finished, #iter = 6192
nu = 0.104976
obj = -604.912230, rho = -0.284038
nSV = 4109, nBSV = 96
......^C

I stopped it because I didn't understand what was going on. I searched online in the libsvm website however I did not find any explanation on why I get multiple outputs and what these outputs mean exactly.

I tried running the heart_scale and I only got one output:

svm-train.exe -t 2 -g 0.07 -c 1 heart_scale
*
optimization finished, #iter = 134
nu = 0.433785
obj = -101.855060, rho = 0.426412
nSV = 130, nBSV = 107
Total nSV = 130

so does the amount of training samples have anything to do with this and if so what's the exact correlation? thanks

Thank you in advance

share|improve this question

1 Answer 1

up vote 1 down vote accepted

Are you doing cross-validation? It's possible that those are just internal debugging ids for the software. I edited my code to not be as verbose, so it usually just prints out summary statistics for me! Based on the fact that it's on the same line as the "optimization finished" statement, it may have to do with iterations while the model is converging. Either way, I don't think you have to worry. Have you read the papers associated with the software? Learning about the underlying algorithm, for me, goes a long way to demystifying it! You should read LIBSVM: A Library for Support Vector Machines. LIBLINEAR: A Library for Large Linear Classification is probably also worth a read!

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.