Calibrating deep convolutional Gaussian processes

Tran, Gia-Lac; Bonilla, Edwin V.; Cunningham, John P.; Michiardi, Pietro; Filippone, Maurizio
AISTATS 2019, 22nd International Conference on
Artificial Intelligence and Statistics, 16-18 April 2019,
Naha, Okinawa, Japan


The wide adoption of Convolutional Neural Networks (CNNs) in applications where decision-making under uncertainty is fundamental, has brought a great deal of attention to the ability of these models to accurately quantify the uncertainty in their predictions. Previous work on combining CNNs with Gaussian processes (GPs) has been developed under the assumption that the predictive probabilities of these models are well-calibrated. In this paper we show that, in fact, current combinations of CNNs and GPs are miscalibrated. We proposes a novel combination that considerably outperforms previous approaches on this aspect, while achieving state-of-the-art performance on image classification tasks.


Type:
Conférence
City:
Naha
Date:
2019-04-15
Department:
Data Science
Eurecom Ref:
5562
Copyright:
© EURECOM. Personal use of this material is permitted. The definitive version of this paper was published in AISTATS 2019, 22nd International Conference on
Artificial Intelligence and Statistics, 16-18 April 2019,
Naha, Okinawa, Japan
 and is available at :

PERMALINK : https://www.eurecom.fr/publication/5562