Exploring the Function Space of Deep-Learning Machines
- Submitting institution
-
Aston University
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 24380717
- Type
- D - Journal article
- DOI
-
10.1103/PhysRevLett.120.248301
- Title of journal
- Physical Review Letters
- Article number
- 248301
- First page
- -
- Volume
- 120
- Issue
- 24
- ISSN
- 0031-9007
- Open access status
- Compliant
- Month of publication
- June
- 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
-
1
- Research group(s)
-
A - Aston Institute of Urban Technology and the Environment (ASTUTE)
- Citation count
- 12
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Engineering achievements of deep-learning machines have intrigued researchers due to our limited understanding of how they work. We provided a general mathematical framework to investigate the function-space of different deep-learning machines based on generating functional analysis. It facilitates studying the capabilities of deep-learning machines and Boolean layered networks and has attracted interest: it was presented in 6 international conferences by invitation (SPML Bangalore 2020, Statistical Physics and Neural Computation Guangzhou 2019, Physics challenges for Machine Learning and Network Science, London 2019, PIL2018 Beijing, PIML2018, Physics Informed Machine Learning Santa-Fe 2018) and resulted in two follow-up papers https://doi.org/10.1088/1751-8121/ab6a6f (invited-featured) and arXiv:2004.08930v1.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
