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...
| Prif Awduron: | , , |
|---|---|
| Fformat: | Conference item |
| Cyhoeddwyd: |
NIPS Foundation
2018
|
| _version_ | 1826264049770299392 |
|---|---|
| author | Gal, Y Hron, J Kendall, A |
| author_facet | Gal, Y Hron, J Kendall, A |
| author_sort | Gal, Y |
| collection | OXFORD |
| description | 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 impossible one with RL. We propose a new dropout variant which gives improved performance and better calibrated uncertainties. Relying on recent developments in Bayesian deep learning, we use a continuous relaxation of dropout’s discrete masks. Together with a principled optimisation objective, this allows for automatic tuning of the dropout probability in large models, and as a result faster experimentation cycles. In RL this allows the agent to adapt its uncertainty dynamically as more data is observed. We analyse the proposed variant extensively on a range of tasks, and give insights into common practice in the field where larger dropout probabilities are often used in deeper model layers. |
| first_indexed | 2024-03-06T20:01:34Z |
| format | Conference item |
| id | oxford-uuid:2779c391-6a38-4c70-be60-7b0c8c88a1a2 |
| institution | University of Oxford |
| last_indexed | 2024-03-06T20:01:34Z |
| publishDate | 2018 |
| publisher | NIPS Foundation |
| record_format | dspace |
| spelling | oxford-uuid:2779c391-6a38-4c70-be60-7b0c8c88a1a22022-03-26T12:07:10ZConcrete dropoutConference itemhttp://purl.org/coar/resource_type/c_5794uuid:2779c391-6a38-4c70-be60-7b0c8c88a1a2Symplectic Elements at OxfordNIPS Foundation2018Gal, YHron, JKendall, ADropout 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 impossible one with RL. We propose a new dropout variant which gives improved performance and better calibrated uncertainties. Relying on recent developments in Bayesian deep learning, we use a continuous relaxation of dropout’s discrete masks. Together with a principled optimisation objective, this allows for automatic tuning of the dropout probability in large models, and as a result faster experimentation cycles. In RL this allows the agent to adapt its uncertainty dynamically as more data is observed. We analyse the proposed variant extensively on a range of tasks, and give insights into common practice in the field where larger dropout probabilities are often used in deeper model layers. |
| spellingShingle | Gal, Y Hron, J Kendall, A Concrete dropout |
| title | Concrete dropout |
| title_full | Concrete dropout |
| title_fullStr | Concrete dropout |
| title_full_unstemmed | Concrete dropout |
| title_short | Concrete dropout |
| title_sort | concrete dropout |
| work_keys_str_mv | AT galy concretedropout AT hronj concretedropout AT kendalla concretedropout |