Ubuntu16.04 Matlab2015+LIBSVM

下载

libsvm下载地址：LIBSM下载，下载libsvm-3.23.tar.gz文件，解压缩，将解压好的文件剪贴到安装好的matlab/toolbox下，例如我的路径是：

/usr/local/MATLAB/R2015b/toolbox

编译安装

在libsvm-3.23所在目录/usr/local/MATLAB/R2015b/toolbox打开终端：

sudo chmod 777 libsvm-3.23
cd libsvm-3.23
make

可以使用./svm-train heart_scale来训练进行检验，会输出如下，并生成heart_scale.model文件：

optimization finished, #iter = 162
nu = 0.431029
obj = -100.877288, rho = 0.424462
nSV = 132, nBSV = 107
Total nSV = 132

接着进行预测：./svm-predict heart_scale heart_scale.model heart_scale.out（其中heart_scale为test file, heart_scale.model是由svm-train训练出来对模型文件，heart_scale.out为预测输出文件），会输出如下，并生成heart_scale.out文件：

Accuracy = 86.6667% (234/270) (classification)

编译完成

与Matlab接口的实现

进入/usr/local/MATLAB/R2015b/toolbox/libsvm-3.23/matlab中，打开终端，make一下

错误一

这个部分忘记截图，错误大概就是mexext not found，主要是想说在libsvm-3.23/matlab中有一个README，里面有写到：

翻译：

在Unix系统上，如果make.m和“mex-setup”都不起作用，请使用Makefile并在命令窗口中键入“make”。
请注意，我们假设您的MATLAB安装在'/usr/local/MATLAB'中。
如果没有，请更换Makefile中的MATLABDIR。

请注意，我们假设您的MATLAB安装在’/usr/local/MATLAB’中。如果没有，请更换Makefile中的MATLABDIR。

• 原始的Makefile文件中
  

而我的Matlab并不是这个安装路径，所以需要更改为自己的MATLAB路径，如下：

错误二

问题：
/usr/local/MATLAB/R2015b/bin/glnxa64/mex: /usr/local/MATLAB/R2015b/bin/glnxa64/libtiff.so.5: no version information available (required by /usr/lib/x86_64-linux-gnu/libopencv_highgui.so.2.4)

libopencv_highgui.so.2.4需要调用 /usr/local/MATLAB/R2015b/bin/glnxa64/libtiff.so.5:。但该版本与libopencv_highgui.so.2.4所能用的版本不对应

• 解决方法：

sudo rm -rf /usr/local/MATLAB/R2015b/bin/glnxa64/libtiff.so.5
 sudo ln -s /usr/lib/x86_64-linux-gnu/libtiff.so.5.2.4 /usr/local/MATLAB/R2015b/bin/glnxa64/libtiff.so.5

linux下Matlab调用及库链接问题no version information available

再重新make一下

/usr/local/MATLAB/R2015b/bin/mex CC="g++" CXX="g++" CFLAGS="-Wall -Wconversion -O3 -fPIC -I/usr/local/MATLAB/R2015b/extern/include -I.." CXXFLAGS="-Wall -Wconversion -O3 -fPIC -I/usr/local/MATLAB/R2015b/extern/include -I.." -largeArrayDims svmpredict.c ../svm.o svm_model_matlab.o
Building with 'gcc'.
Warning: You are using gcc version '5.4.0'. The version of gcc is not supported. The version currently supported with MEX is '4.7.x'. For a list of currently supported compilers see: http://www.mathworks.com/support/compilers/current_release.
Warning: You are using gcc version '5.4.0-6ubuntu1~16.04.12)'. The version of gcc is not supported. The version currently supported with MEX is '4.7.x'. For a list of currently supported compilers see: http://www.mathworks.com/support/compilers/current_release.
/usr/local/MATLAB/R2015b/toolbox/libsvm-3.23/matlab/svmpredict.c: In function ‘int print_null(const char*, ...)’:
/usr/local/MATLAB/R2015b/toolbox/libsvm-3.23/matlab/svmpredict.c:17:36: warning: no return statement in function returning non-void [-Wreturn-type]
 int print_null(const char *s,...) {}
                                    ^

MEX completed successfully.
/usr/local/MATLAB/R2015b/bin/mex CC="g++" CXX="g++" CFLAGS="-Wall -Wconversion -O3 -fPIC -I/usr/local/MATLAB/R2015b/extern/include -I.." CXXFLAGS="-Wall -Wconversion -O3 -fPIC -I/usr/local/MATLAB/R2015b/extern/include -I.." -largeArrayDims svmtrain.c ../svm.o svm_model_matlab.o
Building with 'gcc'.
Warning: You are using gcc version '5.4.0'. The version of gcc is not supported. The version currently supported with MEX is '4.7.x'. For a list of currently supported compilers see: http://www.mathworks.com/support/compilers/current_release.
Warning: You are using gcc version '5.4.0-6ubuntu1~16.04.12)'. The version of gcc is not supported. The version currently supported with MEX is '4.7.x'. For a list of currently supported compilers see: http://www.mathworks.com/support/compilers/current_release.
/usr/local/MATLAB/R2015b/toolbox/libsvm-3.23/matlab/svmtrain.c: In function ‘int read_problem_dense(const mxArray*, const mxArray*)’:
/usr/local/MATLAB/R2015b/toolbox/libsvm-3.23/matlab/svmtrain.c:287:30: warning: conversion to ‘double’ from ‘size_t {aka long unsigned int}’ may alter its value [-Wconversion]
   param.gamma = (double)(1.0/max_index);
                              ^
/usr/local/MATLAB/R2015b/toolbox/libsvm-3.23/matlab/svmtrain.c: In function ‘int read_problem_sparse(const mxArray*, const mxArray*)’:
/usr/local/MATLAB/R2015b/toolbox/libsvm-3.23/matlab/svmtrain.c:370:30: warning: conversion to ‘double’ from ‘size_t {aka long unsigned int}’ may alter its value [-Wconversion]
   param.gamma = (double)(1.0/max_index);
                              ^

MEX completed successfully.
/usr/local/MATLAB/R2015b/bin/mex CC="g++" CXX="g++" CFLAGS="-Wall -Wconversion -O3 -fPIC -I/usr/local/MATLAB/R2015b/extern/include -I.." CXXFLAGS="-Wall -Wconversion -O3 -fPIC -I/usr/local/MATLAB/R2015b/extern/include -I.." -largeArrayDims libsvmread.c
Building with 'gcc'.
Warning: You are using gcc version '5.4.0'. The version of gcc is not supported. The version currently supported with MEX is '4.7.x'. For a list of currently supported compilers see: http://www.mathworks.com/support/compilers/current_release.
Warning: You are using gcc version '5.4.0-6ubuntu1~16.04.12)'. The version of gcc is not supported. The version currently supported with MEX is '4.7.x'. For a list of currently supported compilers see: http://www.mathworks.com/support/compilers/current_release.
MEX completed successfully.
/usr/local/MATLAB/R2015b/bin/mex CC="g++" CXX="g++" CFLAGS="-Wall -Wconversion -O3 -fPIC -I/usr/local/MATLAB/R2015b/extern/include -I.." CXXFLAGS="-Wall -Wconversion -O3 -fPIC -I/usr/local/MATLAB/R2015b/extern/include -I.." -largeArrayDims libsvmwrite.c
Building with 'gcc'.
Warning: You are using gcc version '5.4.0'. The version of gcc is not supported. The version currently supported with MEX is '4.7.x'. For a list of currently supported compilers see: http://www.mathworks.com/support/compilers/current_release.
Warning: You are using gcc version '5.4.0-6ubuntu1~16.04.12)'. The version of gcc is not supported. The version currently supported with MEX is '4.7.x'. For a list of currently supported compilers see: http://www.mathworks.com/support/compilers/current_release.
MEX completed successfully.

添加路径

打开MATLAB，把/usr/local/MATLAB/R2015b/toolbox/libsvm-3.23文件夹路径添加到Matlab的搜索路径（添加并包含子文件夹）
将Matlab的路径设置在/usr/local/MATLAB/R2015b/toolbox/libsvm-3.23下，输入如下测试代码：

[heart_scale_label,heart_scale_inst]=libsvmread('heart_scale');
model = svmtrain(heart_scale_label, heart_scale_inst, '-c 1 -g 0.07');
[predict_label, accuracy, dec_values] =svmpredict(heart_scale_label, heart_scale_inst, model); % test the traindata

显示如下：
完成。

参考：
SVM学习笔记(2)：Ubuntu下LibSVM的安装及与Python和Matlab接口的实现