تقرير
Far-HO: A Bilevel Programming Package for Hyperparameter Optimization and Meta-Learning
العنوان: | Far-HO: A Bilevel Programming Package for Hyperparameter Optimization and Meta-Learning |
---|---|
المؤلفون: | Franceschi, Luca, Grazzi, Riccardo, Pontil, Massimiliano, Salzo, Saverio, Frasconi, Paolo |
سنة النشر: | 2018 |
المجموعة: | Computer Science Statistics |
مصطلحات موضوعية: | Computer Science - Mathematical Software, Computer Science - Learning, Statistics - Machine Learning |
الوصف: | 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. Comment: 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 |
نوع الوثيقة: | Working Paper |
URL الوصول: | http://arxiv.org/abs/1806.04941 |
رقم الأكسشن: | edsarx.1806.04941 |
قاعدة البيانات: | arXiv |
الوصف غير متاح. |