Regularization in Directable Environments with Application to Tetris
- Submitting institution
-
The University of Bath
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 203019145
- Type
- E - Conference contribution
- DOI
-
-
- Title of conference / published proceedings
- Proceedings of Machine Learning Research
- First page
- 3953
- Volume
- 97
- Issue
- -
- ISSN
- 2640-3498
- Open access status
- Compliant
- Month of publication
- June
- Year of publication
- 2019
- URL
-
http://proceedings.mlr.press/v97/lichtenberg19a.html
- Supplementary information
-
https://github.com/janmaltel/stew-tetris
- 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
- -
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Learning from a small number of observations is a difficult problem in machine learning, one that is widely experienced in practice. The paper takes an innovative new approach to this problem, built on two foundations: (1) existing empirical results on the statistical structure of real-world data sets and decision problems, (2) the mathematical framework of regularisation. The approach naturally takes into account a prevalent form of domain knowledge. In addition to showing improvements over existing approaches in supervised learning problems, the paper also demonstrates the technique's potential in reinforcement learning.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -