koraykv/torch-svm

torch-svm

SVM packages for Torch7.

torch-svm/sgd

Reimplementation of Leon Bottou's svmsgd and svmasgd (http://leon.bottou.org/projects/sgd). This implementation is 2-10 times slower depending on the sparsity of the input.

torch-svm/liblinear

This is wrapper around the well known LIBLINEAR library (http://www.csie.ntu.edu.tw/~cjlin/liblinear/).

Requirements

Only Torch7 (http://github.com/torch/torch7)

Building

git clone git://github.com/koraykv/torch-svm.git
luarocks make

Using

require 'svm'

d = svm.ascread('liblinear/liblinear/heart_scale')
model = liblinear.train(d)
labels,accuracy,dec = liblinear.predict(d,model)

Status

• svmsgd and svmasgd implementations are supposed to be complete.
• LIBLINEAR interface is supposed to be complete, I did not check all the possible flags. If you find that something is not working, let me know.
• LIBSVM interface is mostly complete, but as of now passing in a precomputed kernel option is missing.

torch-svm's Issues

Build failure with new split torch

To fix:

1. remove the ADD_TORCH_DOK line from the top-level CMakeLists.txt

2. add lua to dependencies of svmutil, i.e change line 15 in top-level CMakeLists.txt to

   TARGET_LINK_LIBRARIES(svmutil lua luaT TH)

error when building

When I execute "luarocks install svm-0.1-0.rockspec" to implement SVM, then I met the error:
_Using svm-0.1-0.rockspec... switching to 'build' mode

Error: File not found: [email protected]:koraykv/torch-svm.git_

Please help me to fix it. Thank you

Probability estimation using liblinear

Hi,
I'm trying to estimate probabilities using liblinear by passing '-b 1' as an option but getting a nil error.

• - liblinear.predict(test_data,model ['-b 1']) like this. Am I writing the syntax correctly or missing something?

unable to clone into torch-svm

Implementation of scale() function missing.

Nice to have a scale() function for libsmv and liblinear like the original. Would do it but in very early "nob" stage with all this and requires c-programming which i do not know currently. Oh, the technical jargon. LOL