sparseSDCA Version 0.1

C++ code for prox-SDCA

---

0. Changes

This is an old implementation in 2013, which is provided as is. This package has not been updated since.

---

Contents:

1. Introduction

2. System Requirement

3. Installation

4. Examples

5. Contact

6. Copyright

7. References

1. Introduction

This software package provides a sample implementation of accelerated Proximal Stochastic Dual Coordinate Ascent with L1-L2 regularization described in [1] for various loss functions. Please cite the paper if you find the software useful.

The code has not been updated since 2013, and is provided as is.

2. System Requirement

The code has been tested on Linux, but should compile on other unix systems with g++ and make.

git clone https://github.com/TongZhang-ML/sparseSDCA.git

3. Installation

The source files are located in the src directory.

To compile:

cd src/
make

This should compile into two binary programs train and predict

• train: train and save the model;
• predict: apply already trained model on test data.

Use train -h to see command line options for the training program

Use predict -h to see command line options for the prediction program

API documentations are in html/index.html

4. Examples

Please go to the example1 and example2 subdirectories and type run.sh.

5. Contact

[email protected]

6. Copyright

The software is distributed under the MIT license. Please read the file LICENSE.

7. References

[1] Shai Shalev-Shwartz and Tong Zhang. Accelerated Proximal Stochastic Dual Coordinate Ascent for Regularized Loss Minimization, Mathematical Programming, 155:105-145, 2016.

[2] Shai Shalev-Shwartz and Tong Zhang. Stochastic Dual Coordinate Ascent Methods for Regularized Loss Minimization, JMLR 14:567-599, 2013.