Installation Instructions

PHAKISO

YMLL

Microsoft Visual Studio 2005

Unzip all files. All the files should be unzipped into their respective directory. Do not change the directory names.
Add the path containing YMLL.h to Tools -> Options -> Projects and Solutions -> VC++ Directories.
Ensure that the /MD compiler setting is set (Project Property -> C/C++ -> Code Generation -> Runtime Library -> Multi-threaded DLL (/MD)).
Ensure that the project is not using precompiled headers (Project Property -> C/C++ -> Precompiled Headers -> Create/Use Precompiled Header -> Not Using Precompiled Headers).
Ensure that the library file YMLL.lib (and other library files if necessary) is added as input file for the linker (Project Property -> Linker -> Input -> Additional Dependencies -> Full path name of YMLL.lib).
Add #include <YMLL.h> and using namespace YMLL; to source code.

Other Operating System/Compilers

Unzip all files. All the files should be unzipped into their respective directory. Do not change the directory names.
Ensure that the library file libYMLL.a (and other library files if necessary) is added to the compiler options.
Add #include <YMLL.h> and using namespace YMLL; to source code.
If using g++, compile by typing "g++ xxx.cpp -g -w -O -fexceptions -O0 -I path/YMLL/Include -I path/YMLL/Support/galib246 -L path/YMLL/lib -lYMLL". Add more "-l" options if using other libraries, e.g. "-lLibSVM -lSVMLight".

Annie

Needed if using AnnieFFBPNNMachine. Otherwise just comment out the following in YMLL.h.

Download source codes from here.
Unzip all files to a directory.
Create a directory annie and under it, create two directories include and src. The directory annie should be under the YMLL/Support directory.
Copy all header files (.h) to the include directory.
Copy all source code files (.cpp) to the src directory.
Compile into a library file. You can use this perl script to generate a makefile.Annie and then compile using either "make -f makefile.Annie" or "gmake -f makefile.Annie".

galib246

Needed if using GALibDescriptorSelection or GALibGroupDescriptorSelection. Otherwise just comment out the following in YMLL.h.

Download source codes from here.
Unzip all files to a directory.
The whole directory galib246 should be under the YMLL/Support directory. If a newer version of galib is used, the #include statement in DescriptorSelectionGALib.h and DescriptorSelectionGALibGroup.h should be modified accordingly.
Compile into a library file. You can use this perl script to generate a makefile.galib246 and then compile using either "make -f makefile.galib246" or "gmake -f makefile.galib246". (Note: for Visual Studio 2005, you may have to rename all .c files to .cpp files.)

LibSVM

Needed if using LibSVMMachine. Otherwise just comment out the following in YMLL.h.

Download source codes from here.
Unzip all files to a directory LibSVM which should be created under the YMLL/Support directory.
Rename svm.h and svm.cpp to libsvm.h and libsvm.cpp respectively.
Change #include "svm.h" in libsvm.cpp to #include "libsvm.h".
Move struct svm_model and struct decision_function from libsvm.cpp to libsvm.h.
Insert function declaration for svm_train_one, svm_svr_probability, svm_group_classes and svm_binary_svc_probability into libsvm.h.
Compile into a library file. You can use this perl script to generate a makefile.LibSVM then compile using either "make -f makefile.LibSVM" or "gmake -f makefile.LibSVM".

PNN

Needed if using MasterPNNMachine, MasterPNNSepClassMachine or MasterGRNNMachine. Otherwise just comment out the following in YMLL.h.

Get source codes from Masters T (1995). Advanced algorithms for neural networks : a C++ sourcebook. New York, Wiley.
Copy all files to a directory PNN which should be created under the YMLL/Support directory.
Compile into a library file. You can use this perl script to generate a makefile.PNN and then compile using either "make -f makefile.PNN" or "gmake -f makefile.PNN".

SVM^light

Needed if using SVMLightMachine. Otherwise just comment out the following in YMLL.h.

Download source codes from here.
Unzip all files to a directory SVMLight which should be created under the YMLL/Support directory.
For Visual Studio 2005, convert all WORD to SVMWORD because it conflicts with Windows typedef SVMWORD.
Compile into a library file. You can use this makefile: makefile.SVMLight and then compile using either "make -f makefile.SVMLight" or "gmake -f makefile.SVMLight".

Torch

Needed if using TorchSVMMachine or TorchMLPMachine. Otherwise just comment out the following in YMLL.h.

Download source codes from here.
Unzip all files to a directory Torch which should be created under the YMLL/Support directory.
Compile into a library file. You can use this perl script to generate a makefile.Torch and then compile using either "make -f makefile.Torch" or "gmake -f makefile.Torch".