PyHubs is a machine learning library developed in Python. It contains implementations of
hubness-aware machine learning algorithms together with some useful tools for machine learning experiments.
In particular, PyHubs
- contains implementations of hubness-aware classifiers and
- hubness-aware regression techniques,
- supports semi-supervised classification according to the self-training schema,
- supports evaluation of classification and regression techniques: the module contains methods implementing popular experimental protocols, and
The software, together with its documentation and examples illustrating how to use it, can be downloaded from here.
You may also have a look at our publications about hubness-aware machine learning approaches.
According to our recent observation, old versions of PyHubs (such as 1.2.1) does not provide correct results with new versions
of numpy (such as 1.16), however, we think that the most recent version of PyHubs (1.3) works correctly with new versions of numpy as well.