1. Core Features
2. Installation Guide
3. Quick Start
4. Documentation
5. Development
6. Donations
7. Citing LIBTwinSVM
8. License

LIBTwinSVM is an easy-to-use implementation of Twin Support Vector Machine. It is licensed under the terms of GNU GPL v3. This application comes with a user interface which makes using the algorithm so handy for Data Scientists, Machine Learning Researchers and whoever else that is interested in Machine Learning.

• A simple and user-friendly Graphical User Interface (GUI).
• Supports both standard TwinSVM and Least Squares TwinSVM classifiers.
• Easy to import data in CSV & LIBSVM format.
• A dataset can be loaded with shuffling and normalization.
• A fast optimizer (clipDCD) is improved and implemented in C++ to solve optimization problems of TwinSVMs.
• Supports Linear, RBF and Rectangular kernel.
• Supports Binary and Multi-class classification (One-vs-All & One-vs-One).
• The OVA and OVO estimators are compatible with scikit-learn tools such as GridSearchCV, cross_val_score, etc.
• A classifier can be evaluated using either K-fold cross-validation or Training/Test split.
• Supports grid search over estimators' hyper-parameters.
• The detailed classification results can be saved in an Excel-format spreadsheet file.
• The classification results can be logged during the grid search process to not lose results in case of power failure.
• A feature-rich visualization tool to show decision boundaries and geometrical interpretation of TwinSVMs.
• The best-fitted classifier can be saved on the disk after the grid search process.
• The pre-trained models can be loaded and evaluated on the test samples.

LIBTwinSVM depends on the following packages.

For Installing LIBTwinSVM you can choose one of the following commands based on your Operating System and your Python version. Please note that for installing the latest bugfixes and features, we recommend you to install the library from the source.

• Linux & Mac OS:

pip3 install libtwinsvm

• Windows:

pip install libtwinsvm

1. Downloading LIBTwinSVM

First, make sure that Git is installed as it is required for getting the source code. Then open Git in any arbitrary path and enter the following command:

git clone --recursive https://github.com/mir-am/LIBTwinSVM

2. Downloading Requirements

Before installing any packages, we recommend you to upgrade pip:

• Linux & Mac OS:

pip3 install -U pip

• Windows:

pip install -U pip

For LIBTwinSVM installation, Numpy and Cython must be installed. You can install them by entering the following commands in your terminal. If you already have Cython and Numpy installed, you can skip to the next section.

• Linux & Mac OS: For installing Numpy & Cython on your computer, you should enter the following command in the terminal.

pip3 install numpy cython

Note for Linux users: You should also install the following packages for your distribution:

Debian-based Linux distro:

sudo apt-get install python3-tk liblapack-dev libblas-dev

RPM-based Linux distro:

sudo yum install python3-tkinter lapack-devel blas-devel

• Windows: for installing the requirements on windows, you should enter the following command in the command prompt.

pip install numpy cython

3. Installing LIBTwinSVM

Go to LIBTwinSVM folder where you have downloaded the source code in step 2. Then enter the following command in the terminal:

• Linux & Mac OS:

pip3 install .

• Windows:

pip install .

For uninstalling LIBTwinSVM, enter the following command in the terminal:

• Linux & Mac OS:

pip3 uninstall libtsvm

• Windows:

pip uninstall libtsvm

After LIBTwinSVM was installed, for running the GUI application, open the terminal and enter the following code:

• Linux & Mac OS:

python3 -m libtsvm

• Windows:

python -m libtsvm

LIBTwinSVM supports data files with the following formats:

• CSV files with '.csv' extention.
• LIBSVM files with '.libsvm' extention.

Note that for importing data as a CSV file, the dataset has to comply the following terms:

1. The first column has to be the dataset labels. Moreover, labels of positive and negative samples should be 1 and -1, respectively.
2. The first row can be dataset's headernames. (Optional)
3. All the values in dataset except headernames should be numerical. Nominal values are not allowed.
   

To help you prepare your dataset and test the program, three datasets are included here.

Usage examples and API reference can be found on the project's Read the Docs page.

After installing the library, you can run unit tests to ensure that your installation is valid. To do so, you need to install pytest package with the following command:

pip install pytest

Next, execute the following command in the root of the project directory:

pytest tests

If you have used the LIBTwinSVM project and found it helpful, please consider making a donation via PayPal to support this work. It also motivates us to maintain the project and develop new features.

If you have used the LIBTwinSVM library in your research work, please cite the following paper:

• Mir, A. M., Rahbar, M., & Nasiri, J. A. (2020). LIBTwinSVM: A Library for Twin Support Vector Machines. arXiv preprint arXiv:2001.10073.

BibTeX entry:

@article{mir2020libtwinsvm,
  title={LIBTwinSVM: A Library for Twin Support Vector Machines},
  author={Mir, Amir M and Rahbar, Mahdi and Nasiri, Jalal A},
  journal={arXiv preprint arXiv:2001.10073},
  year={2020}
}

LIBTwinSVM library is licensed under the terms of GNU General Public License v3. This library can be used for both academic and commercial purposes. For more information, check out the LICENSE file.