Conference site ยป Proceedings

Hyperopt-Sklearn: Automatic Hyperparameter Configuration for Scikit-Learn

Brent Komer
Centre for Theoretical Neuroscience, University of Waterloo

James Bergstra
Centre for Theoretical Neuroscience, University of Waterloo

Chris Eliasmith
Centre for Theoretical Neuroscience, University of Waterloo

Abstract

Hyperopt-sklearn is a new software project that provides automatic algorithm configuration of the Scikit-learn machine learning library. Following Auto-Weka, we take the view that the choice of classifier and even the choice of preprocessing module can be taken together to represent a single large hyperparameter optimization problem. We use Hyperopt to define a search space that encompasses many standard components (e.g. SVM, RF, KNN, PCA, TFIDF) and common patterns of composing them together. We demonstrate, using search algorithms in Hyperopt and standard benchmarking data sets (MNIST, 20-Newsgroups, Convex Shapes), that searching this space is practical and effective. In particular, we improve on best-known scores for the model space for both MNIST and Convex Shapes.

Keywords

bayesian optimization, model selection, hyperparameter optimization, scikit-learn

Bibtex entry

Full text PDF