In (Franceschi et al., 2018) we proposed a unified mathematical framework, grounded on bilevel programming, that encompasses gradient-based hyperparameter optimization and meta-learning. We formulated an approximate version of the problem where the inner objective is solved iteratively, and gave sufficient conditions ensuring convergence to the exact problem. In this work we show how to optimize learning rates, automatically weight the loss of single examples and learn hyper-representations with Far-HO, a software package based on the popular deep learning framework TensorFlow that allows to seamlessly tackle both HO and ML problems.
@article{arxiv.1806.04941,
title = {Far-HO: A Bilevel Programming Package for Hyperparameter Optimization and Meta-Learning},
author = {Luca Franceschi and Riccardo Grazzi and Massimiliano Pontil and Saverio Salzo and Paolo Frasconi},
journal= {arXiv preprint arXiv:1806.04941},
year = {2018}
}
Comments
This submission is a reduced version of (Franceschi et al., arXiv:1806.04910) which has been accepted at the main ICML 2018 conference. In this paper we illustrate the software framework, material that could not be included in the conference paper