tiny little machine learning toolbox
Tilitools is basically a collection of (non-mainstream) machine learning model and tools with a special focus on anomaly detection, one-class learning, and structured data. Furthermore, we emphasize simplicity and ease-of-use not runtime performance (although we put some effort into optimization). Descriptive examples can be found in the notebooks/ and scripts/ sub-directories.
written by Nico Goernitz
Currently available models:
-
(dual) support vector data description
-
(dual) one-class support vector machine
-
(convex,dual) semi-supervised anomaly detection method
-
corresponding lp-norm mkl wrapper
-
(primal) structured output support vector machine
-
(primal) non-convex latent variable support vector data description
-
corresponding multi-class joint feature map and methods
-
examples