artlearn is a set of algorithms for Adoptive resonance theory.
This contains these techniques.
- ART1
- ART2
- ART2A
- Bayesian ART
- Fuzzy ART
- SFAM
and my original,
- L2ART
The required dependencies to use artlearn are
- scikit-learn >= 0.22.2
- numpy >= 1.19.5
- scipy >= 1.4.1
You also need Matplotlib >= 3.2.2 to run the demo and pytest >= 6.2.5 to run the tests.
pip install artlearnWe have posted a usage example in the demo folder.
For example...
from artlearn import FuzzyART
clf = FuzzyART(max_iter=100, max_class=100, rho=0.72, alpha=1e-5, beta=0.1)
clf.fit(X)
labels = clf.labels_from artlearn import BayesianART
clf = BayesianART(max_iter=3, max_class=100, rho=0.01, sigma=0.05, max_hyper_volume=0.07)
clf.fit(X)
labels = clf.labels_from artlearn import SFAM
clf = SFAM(max_iter=100, max_class=100, rho=0.9, alpha=1e-5, beta=0.1)
clf.fit(X, y)
y_pred = clf.predict(X)This code is licensed under MIT License.
python setup.py test