A theoretically grounded application of dropout in recurrent neural networks
- Submitting institution
-
University of Oxford
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 2094
- Type
- E - Conference contribution
- DOI
-
-
- Title of conference / published proceedings
- Advances in Neural Information Processing Systems 29 (NIPS 2016)
- First page
- 1019
- Volume
- 29
- Issue
- -
- ISSN
- 1049-5258
- Open access status
- Compliant
- Month of publication
- December
- Year of publication
- 2016
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
1
- Research group(s)
-
-
- Citation count
- 10
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- The tool developed in this paper is implemented as a default in all major deep learning software packages, including Tensorflow, Pytorch, Keras, and DeepMind's codebase. As of 2016, it has effectively been used in industry and academia by almost all AI models which parse sequences of text or images. The tool is used in many state-of-the-art models in natural language applications such as in machine translation (Birch et al., WMT’16: https://www.aclweb.org/anthology/W16-2323/); it is also used in many follow-up influential works in the field (e.g. Zoph and Le, Google, 2016: https://arxiv.org/abs/1611.01578), and in work published at Nature (https://doi.org/10.1038/s41598-018-24271-9).
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -