Concrete dropout
Dropout is used as a practical tool to obtain uncertainty estimates in large vision models and reinforcement learning (RL) tasks. But to obtain well-calibrated uncertainty estimates, a grid-search over the dropout probabilities is necessary— a prohibitive operation with large models, and an impossib...
主要な著者: | Gal, Y, Hron, J, Kendall, A |
---|---|
フォーマット: | Conference item |
出版事項: |
NIPS Foundation
2018
|
類似資料
-
Dropout inference in Bayesian neural networks with alpha-divergences
著者:: Li, Y, 等
出版事項: (2017) -
A theoretically grounded application of dropout in recurrent neural networks
著者:: Gal, Y, 等
出版事項: (2016) -
Dropout in post-socialist countries of Central Europe (the Visegrad Group)
著者:: Petr Gal
出版事項: (2021-05-01) -
Dropout intention: a valid predictor of actual dropout?
著者:: Stefanie Findeisen, 等
出版事項: (2024-06-01) -
Dropout rates and reasons for dropout among patients receiving clozapine
著者:: Sandeep Grover, 等
出版事項: (2023-06-01)