SEFR is a fast, linear-time classifier to be used mainly on, but not limited to, ultra-low power devices. The data should be normalized in the range [0, n], and its labels should be 0 and 1 (this will be fixed in the next version). For training and prediction, fit() and predict() methods are used, the same way as in other classifiers.
hamidmandi / sefr Goto Github PK
View Code? Open in Web Editor NEWThis project forked from sefr-classifier/sefr
The SEFR Classifier
License: MIT License