In an orchestra, the oboe plays an initial note which the other instruments use to tune to the right frequency before the performance begins; this package, Oboe, is an automated machine learning/model selection system that uses collaborative filtering to find good models for supervised learning tasks within a user-specified time limit. Further hyperparameter tuning can be performed afterwards.
Oboe is based on matrix factorization and classical experiment design. For a complete description, refer to our paper at KDD 2019: OBOE: Collaborative Filtering for AutoML Model Selection.
This system is still under developement and subjects to change.
The following packages/libraries are required. The versions in brackets are the versions that are verified to work; a higher (or lower) version may still works.
- Python (3.7.3)
- numpy (1.16.4)
- scipy (1.4.1)
- pandas (0.24.2)
- scikit-learn (0.22.1)
- multiprocessing (>=0.70.5)
- OpenML (0.9.0)
- mkl (>=1.0.0)
- re
- os
- json
- tensorly
This part is currently under development; an example for code usage is in the example
folder. The package will be pip installable in the future.
Updating ...