Metadata-Version: 1.0
Name: Lurrn
Version: 0.7.3
Summary: Simple machine learning library
Home-page: UNKNOWN
Author: Yannick Versley
Author-email: yversley@gmail.com
License: UNKNOWN
Description: Lurrn
        =====
        
        Lurrn is a library for online learning of linear classifiers which can be
        used to learn, e.g. CRFs. It's mostly meant as an educational tool for
        demonstrating structured learning, but could also be used for more serious
        undertakings.
        
        Basic usage
        -----------
        
        For a medium complexity example, see examples/linear_crf.py.
        
        The basic idea is: you represent your example as a list of strings,
        use a FeatureHasher to turn this into a (sparse) feature vector,
        and can then hand it to a learner to get its score wrt. a weight
        vector (learner.score(vec)), or to update the weight vector with
        this gradient (learner.update([(1.0, vec)]). Using the SparseVectorsD
        class, you can also combine multiple vectors for a gradient update.
        
Keywords: machine learning,numpy
Platform: UNKNOWN
