Python implementations of metric learning algorithms
Metric Learning algorithms in Python.
- Large Margin Nearest Neighbor (LMNN)
- Information Theoretic Metric Learning (ITML)
- Sparse Determinant Metric Learning (SDML)
- Least Squares Metric Learning (LSML)
- Neighborhood Components Analysis (NCA)
- Local Fisher Discriminant Analysis (LFDA)
- Relative Components Analysis (RCA)
- Python 2.7+, 3.4+
- numpy, scipy, scikit-learn
- (for running the examples only: matplotlib)
Run pip install metric-learn to download and install from PyPI.
Run python setup.py install for default installation.
Run python setup.py test to run all tests.
For full usage examples, see the sphinx documentation.
Each metric is a subclass of BaseMetricLearner, which provides default implementations for the methods metric, transformer, and transform. Subclasses must provide an implementation for either metric or transformer.
For an instance of a metric learner named foo learning from a set of d-dimensional points, foo.metric() returns a d by d matrix M such that a distance between vectors x and y is expressed (x-y).dot(M).dot(x-y).
In the same scenario, foo.transformer() returns a d by d matrix L such that a vector x can be represented in the learned space as the vector x.dot(L.T).
For convenience, the function foo.transform(X) is provided for converting a matrix of points (X) into the learned space, in which standard Euclidean distance can be used.
If a recent version of the Shogun Python modular (modshogun) library is available, the LMNN implementation will use the fast C++ version from there. The two implementations differ slightly, and the C++ version is more complete.