Comments on: Support Vector Machines (SVM) in Ruby http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/ A goal is a dream with a deadline. Sat, 30 Aug 2008 00:15:41 +0000 http://wordpress.org/?v=2.5.1 By: Using Ruby for Bioinformatics Applications at Bleeding Edge Biotech http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-103023 Using Ruby for Bioinformatics Applications at Bleeding Edge Biotech Tue, 06 May 2008 16:07:38 +0000 http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-103023 [...] Support Vector Machines in Ruby Fast Artificial Neural Network library [...] [...] Support Vector Machines in Ruby Fast Artificial Neural Network library [...]

]]> By: Ben http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-102927 Ben Sat, 03 May 2008 08:52:04 +0000 http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-102927 Excellent article. Thanks! Excellent article. Thanks!

]]> By: sean http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-102083 sean Mon, 17 Mar 2008 02:51:03 +0000 http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-102083 What, no discussion of stemming? I would hope that nobody goes about implementing such a system without looking into normalizing and stemming the bag of words before creating a dictionary! What, no discussion of stemming? I would hope that nobody goes about implementing such a system without looking into normalizing and stemming the bag of words before creating a dictionary!

]]> By: Richard heycock http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-101623 Richard heycock Sat, 23 Feb 2008 04:18:01 +0000 http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-101623 Hey Jason, The snowball project has a good stop word list: http://snowball.tartarus.org/dist/snowball_all.tgz It's got lists for a fair few European languages as well. Hey Jason,

The snowball project has a good stop word list:
http://snowball.tartarus.org/dist/snowball_all.tgz

It’s got lists for a fair few European languages as well.

]]> By: Ilya Grigorik http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-101578 Ilya Grigorik Tue, 19 Feb 2008 03:21:28 +0000 http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-101578 Jason, libsvm supports multi-category classification right out of the box! Simply assign different labels to your training examples and you're all ready to go. For more info, check <a href="http://www.csie.ntu.edu.tw/%7Ecjlin/libsvm/faq.html" rel="nofollow">LibSVM FAQ</a>. Jason, libsvm supports multi-category classification right out of the box! Simply assign different labels to your training examples and you’re all ready to go. For more info, check LibSVM FAQ.

]]> By: Jason F http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-101535 Jason F Thu, 14 Feb 2008 18:45:45 +0000 http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-101535 For some reason my email didn't show up in my last comment - it's rormaniac@mac.com For some reason my email didn’t show up in my last comment - it’s rormaniac@mac.com

]]> By: Jason F http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-101534 Jason F Thu, 14 Feb 2008 18:44:44 +0000 http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-101534 How would modify this code to support multiple categories, instead of just two (spam/notspam)? Is it just a matter of adding more arrays with different content? How many categories can an SVM handle well? Also, if anyone has a good stopword list, please let me know- thanks! How would modify this code to support multiple categories, instead of just two (spam/notspam)? Is it just a matter of adding more arrays with different content? How many categories can an SVM handle well?

Also, if anyone has a good stopword list, please let me know- thanks!

]]> By: HG http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-101485 HG Tue, 12 Feb 2008 14:05:28 +0000 http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-101485 Ilya, sorry for turning your comments into a forum, but if I may: comment to Kaven => Your problem has nothing to do with the bindings. Try serializing the objects with, for example, the ruby yaml library. #create model as in Ilya's code: m = Model.new(sp, pa) open("your_model.yml", 'w') { |f| YAML.dump(m, f) } #Load the model a month from now: m = open("your_model.yml", 'r') { |f| YAML.load(f) } #and make a predicion pred, probs = m.predict_probability(test_vectors[i]) Off topic: I still haven't been able to compile the ruby bindings on windows xp. If anyone out there can shed some light! ;) Ilya, sorry for turning your comments into a forum, but if I may: comment to Kaven =>
Your problem has nothing to do with the bindings. Try serializing the objects with, for example, the ruby yaml library.

#create model as in Ilya’s code:
m = Model.new(sp, pa)
open(”your_model.yml”, ‘w’) { |f| YAML.dump(m, f) }
#Load the model a month from now:
m = open(”your_model.yml”, ‘r’) { |f| YAML.load(f) }
#and make a predicion
pred, probs = m.predict_probability(test_vectors[i])

Off topic: I still haven’t been able to compile the ruby bindings on windows xp. If anyone out there can shed some light! ;)

]]> By: Ilya Grigorik http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-101479 Ilya Grigorik Tue, 12 Feb 2008 11:51:16 +0000 http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-101479 Kaven, there's the problem! We're talking about two different bindings: ruby vs python. Of course, I'm referring to the ruby bindings, and I don't know if the python ones come with all the same functionality -- they are implemented by different people. I would suggest digging into the docs for your binding, or contacting the author(s)! Kaven, there’s the problem! We’re talking about two different bindings: ruby vs python. Of course, I’m referring to the ruby bindings, and I don’t know if the python ones come with all the same functionality — they are implemented by different people. I would suggest digging into the docs for your binding, or contacting the author(s)!

]]> By: kaven http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-101434 kaven Mon, 11 Feb 2008 06:57:02 +0000 http://www.igvita.com/2008/01/07/support-vector-machines-svm-in-ruby/#comment-101434 Thanks for your comment and great series about ai and ml. I read test.rb. But I can't find methods to save and load models. In "libsvm/python/svm.py", save and load is : svmc.svm_save_model(filename,self.model) svmc.svm_save_model(filename,self.model). Which methods to save and load in libsvm-ruby? Thanks for your comment and great series about ai and ml.
I read test.rb. But I can’t find methods to save and load models. In “libsvm/python/svm.py”, save and load is : svmc.svm_save_model(filename,self.model)
svmc.svm_save_model(filename,self.model).
Which methods to save and load in libsvm-ruby?

]]>