Learnable Bernoulli Dropout for Bayesian Deep Learning

Boluki Shahin
Boluki Shahin
Ardywibowo Randy
Ardywibowo Randy
Dadaneh Siamak Zamani
Dadaneh Siamak Zamani

AISTATS, pp. 3905-3916, 2020.

Cited by: 5|Views32
EI

Abstract:

In this work, we propose learnable Bernoulli dropout (LBD), a new model-agnostic dropout scheme that considers the dropout rates as parameters jointly optimized with other model parameters. By probabilistic modeling of Bernoulli dropout, our method enables more robust prediction and uncertainty quantification in deep models. Especially,...More

Code:

Data:

Your rating :
0

 

Tags
Comments