On May 4, 12:10 pm, Martin Albrecht <[email protected]> wrote: > On Tuesday 04 May 2010, William Stein wrote: > > > > > > > On Tue, May 4, 2010 at 11:11 AM, William Cauchois <[email protected]> > wrote: > > > Hi everyone, > > > > As part of another project, I completed a rudimentary Python binding > > > to Thorsten Joachims' SVM-Light library (http:// > > > svmlight.joachims.org/) implementing a Support Vector Machine. My > > > source code is available at > > > >http://bitbucket.org/wcauchois/pysvmlight > > > > Are there people doing ML research who would be interested in having > > > access to SVM-Light from inside Sage? Is there anyone who would like > > > to help me get PySVMLight into a more feature-complete state and then > > > integrate it with Sage? > > > Could you write a little more to sage-devel about why mathematicians > > might care about "support vector machines" -- it's possible that most > > people reading this have never heard of them. > > Since I'm nowhere near an expert, here's the first paragraph from Wikipedia: > > """ > Support vector machines (SVMs) are a set of related supervised learning > methods used for classification and regression. In simple words, given a set > of training examples, each marked as belonging to one of two categories, an > SVM training algorithm builds a model that predicts whether a new example > falls into one category or the other. Intuitively, an SVM model is a > representation of the examples as points in space, mapped so that the examples > of the separate categories are divided by a clear gap that is as wide as > possible. New examples are then mapped into that same space and predicted to > belong to a category based on which side of the gap they fall on. > """ > > http://en.wikipedia.org/wiki/Support_vector_machine > > I recently played around with SVM in the context of cryptanalysis to > distinguish plaintext-ciphertext pairs with certain "good" properties from > others as part of an attack. For that I actually wrote a little Sage script :) > I assume SVMs could be useful to classify experimental data etc. in other > contexts too. > > Btw. I came across SVMs in the context of author identification: you extract > features such as sentence length and then use those feature vectors to decide > whether an anonymous text was written by A or B. The idea is that you do not > need to build a model yourself, the algorithm will attempt to separate your > training sets as much as possible. > > I'd like to see some optional SPKG which provides SVM support. However, in the > foreseeable future I won't be able to help much since I'm supposed to work on > my thesis. > > Cheers, > Martin
Thanks for sharing your experiences. An optional SPKG sounds like the way to go. Could you point me to any resources describing how to go about creating an SPKG? Another thing I'm not clear on is how to integrate a Python C extension with Sage. There must be some SPKGs that do this, so maybe I could look at that source code. -- To post to this group, send an email to [email protected] To unsubscribe from this group, send an email to [email protected] For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
