Diversity regularization in deep ensembles

Autor:	Shui, Changjian, Mozafari, Azadeh Sadat, Marek, Jonathan, Hedhli, Ihsen, Gagné, Christian
Rok vydání:	2018
Předmět:	Computer Science - Learning
Druh dokumentu:	Working Paper
Popis:	Calibrating the confidence of supervised learning models is important for a variety of contexts where the certainty over predictions should be reliable. However, it has been reported that deep neural network models are often too poorly calibrated for achieving complex tasks requiring reliable uncertainty estimates in their prediction. In this work, we are proposing a strategy for training deep ensembles with a diversity function regularization, which improves the calibration property while maintaining a similar prediction accuracy. Comment: 6 pages
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/1802.07881 Zobrazit plný text záznamu View this record from Arxiv