This repository contains a C++ implementation and a Python wrapper for FRiS-STOLP classifier.
The algorithm's main idea is to find a subset of representatives in each class of the training set. These representatives are then used to approximate the borderline between the classes using the function of rival similarity (FRiS).
- Boost.uBLAS
- Boost.Python
- Binary classifier
- Memoized and non-memoized distance calculation
- Python wrapper
- sklearn adapter without parameter importance estimation
- Full sklearn compatibility
- Automatic installation