Breaking the Activation Function Bottleneck through Adaptive Parameterization
- Submitting institution
-
The University of Manchester
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 85439236
- Type
- E - Conference contribution
- DOI
-
-
- Title of conference / published proceedings
- Advances in Neural Information Processing Systems 31 (NeurIPS 2018)
- First page
- 0
- Volume
- -
- Issue
- -
- ISSN
- -
- Open access status
- -
- Month of publication
- November
- Year of publication
- 2018
- 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
-
3
- Research group(s)
-
A - Computer Science
- Citation count
- 0
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- "The new method for neural networks demonstrates a significant increase in model expressiveness for a given parameter count.
Invited talk to present the method at Amazon CoreML (Cambridge 2018).
Resulted in multiple, high level placements for the PGR (Flennerhag) ATI enrichment scheme, 9 months; Amazon internship - 5 months; DeepMind internship - 6 months, leading to full-time position at DeepMind (May 2020).
Take up on social media: https://twitter.com/Miles_Brundage/status/999100623856586752?s=20
Work available in github https://github.com/flennerhag/alstm (August 2020, ~100 downloads)."
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
