Overview¶
GraKeL is a Python package which provides implementations of several graph kernels, a family of powerful methods which allow kernel-based learning approaches such as SVMs to work directly on graphs.
Getting Started
Benchmarks¶
To demonstrate the efficiency of the algorithms implemented in GraKeL, we present a comparison of the running times of the implementations of some graph kernels from GraKeL and from other packages. We also compare the running times of the different kernels to each other.
Package Reference¶
A collection of all classes and functions important for the use and understanding of the GraKeL package.
GraKeL provides
What’s New¶
Version 0.1a7
Detailed installation instructions for c++ extensions in windows.
Changed base_kernel alias in frameworks with base_graph_kernel to disambiguate with vectorial kernels.
Speed-up for floyd_warshall calculation in graph.py.
Large update throughout all the documentation
Version 0.1a6
More scikit-learn compatibility:
Initialise kernels by name and alias on GraphKernel (as GraphKernel(kernel=”shortest_path”).
Fit and instantion by default parameters.
Random number generator standardized check_random_state. random_seed are now random_state arguments.
Doctests.
Miscelanous:
Detailed unsupported kernel output.
More detailed licensing information considering cvxopt and BLISS
Small bugfix inside the (Count Sensitive) Neighborhood Hash Kernel.
Added sparse-compatibility for VertexHistogram and for EdgeHistogram.
Version 0.1a5
Various bugfixes in kernel implementations.
Added a bunch of
utils
functions for external operations: transforming existing graph formats (csv, pandas, networkx) to the grakel native, k-fold cross validation with an SVM and kernel matrix transformer for manipulating precomputed kernel matrices in anTransformer
fashion.Conda compatibility: visit https://anaconda.org/ysig/grakel-dev.