All geek questions in one place

Need an SVM implementation or Java library

I have a data set with 2400 samples and 10,000 functions. All data is binary (+1 or -1). I need to run it through the SVM algorithm so that I can compare my algorithm with it. However, I know little about SVM or which package to use. I tried to read so that I could implement one, but this is on my head, and I need to get a weight vector from it. I am a Windows user and I have a Java implementation. I could export my data to a text file with 1 and 0. I have access to MatLab, but something tells me that it will be very slow and actually will not work on my laptop with a capacity of 6 GB (and this is for me need) fast enough. I have to run the algorithm a couple of hundred times to get accurate results.

I'm really looking for a quick and easy understanding of the library or SVM implementation that I can use in my case.

Thanks to everyone. Feel free to ask additional questions to help me better.

+3
source share
3 answers

I ended up using JNI for SVN light, which can be found here: http://www.mpi-inf.mpg.de/~mtb/svmlight/

It did not take long to figure out how to use it, and it is amazingly fast (seconds).

+4
source

I don’t think that there is any way to "quick and easy understanding of SVM". The math is hard and trying to train one without a good understanding is a very quick trip to shoot yourself in the foot.

OpenSVM sourceforge, , . , Java, JDK.

SVM MatLab, , - .

+2

I am a student who is trying to learn SVM. If you still have code, can you pass it ?? !! this is part of my cool project.

0
source

Source: https://habr.com/ru/post/1766591/

More articles:


All Articles