A collection of python code to perform research in optimization. The aim is to provide reusable components that can be quickly applied to machine learning problems.
- python 2.5+
- cvxopt 1.0+ (for solving linear and quadratic programs)
- pythongrid (for using a cluster)
- cython 0.14.1 (for speeding up kernel computations)
- Download OptWok-0.3.1.tar.gz.
Release notes (Optwok-0.3.1):
- Minor bugfix release
- Download OptWok-0.3.tar.gz.
Release notes (Optwok-0.3):
- Implemented Ellipsoidal Multiple Instance Learning
- Included code for Gaussian Process Contextual Bandits
- difference of convex functions algorithms for sparse classfication
- Download OptWok-0.2.1.tar.gz.
Release notes (OptWok-0.2.1):
- Fixed missing multiclass module.
- Slycot sources no longer distributed, using github project instead.
Release notes (OptWok-0.2):
- Use block coordinate descent to learn the kernel on outputs.
- Download OptWok-0.1.tar.gz.
Release notes (OptWok-0.1):
- Initial release
- Implements 8 methods, which is the combination of 2 losses and 4 regulariers. 6 are DCA methods.
- hinge loss
- ramp loss
- l2, l1, l0, capped l1
- Ellipsoidal Multiple Instance Learning. (webpage)
- Learning sparse classifiers with Difference of Convex functions Algorithms, (journal)
- Contextual Gaussian Process Bandit Optimization.
- Learning Output Kernels with Block Coordinate Descent (webpage)
- Learning sparse classifiers with difference of convex functions algorithms (webpage)