Symmetry-based disentangled representation learning requires interaction with environments
- Submitting institution
-
City, University of London
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 822
- Type
- E - Conference contribution
- DOI
-
-
- Title of conference / published proceedings
- Advances in Neural Information Processing Systems
- First page
- 1
- Volume
- 32
- Issue
- -
- ISSN
- 1049-5258
- Open access status
- Compliant
- Month of publication
- December
- Year of publication
- 2019
- URL
-
https://papers.nips.cc/paper/2019/hash/36e729ec173b94133d8fa552e4029f8b-Abstract.html
- 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
-
2
- Research group(s)
-
-
- Citation count
- 0
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- This paper appeared in NeurIPS 2019, the leading conference in AI (13000 attendees, 21.6 % acceptance rate). It introduces a mathematical framework that formalises sensorimotor representation learning for naïve agents and shows that such learning significantly improves discovery of spatial properties of the world. This work was cited in papers from prominent international research teams (Carnegie Mellon University, DeepMind, INRIA, ENS Paris, University of Oxford, NNAISSENCE, UC Berkeley). It was also key to the success of a Dstl Grant application (k£98) that finances a PhD student co-supervised with ENSTA Paris.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
