Efficient Gradient-Free Variational Inference using Policy Search
- Submitting institution
-
University of Aberdeen
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 171264631
- Type
- E - Conference contribution
- DOI
-
-
- Title of conference / published proceedings
- Proceedings of the 35th International Conference on Machine Learning
- First page
- 234
- Volume
- -
- Issue
- -
- ISSN
- 2640-3498
- Open access status
- Technical exception
- Month of publication
- July
- 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
-
2
- Research group(s)
-
-
- Citation count
- -
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- This paper proposes a novel generic gradient-free variational inference algorithm for employing a mixture of Gaussian models to approximate both multimodal and intractable distributions when only distribution samples are available. This is original and significant especially for the robotics community because the method is able to explore and exploit the complex distributions commonly arisen in problems of that domain which are not only multimodal but also intractable, i.e., the target distributions have no closed forms and thus gradients are not available. The method has been adapted or used in robotics applications (https://tinyurl.com/y4rnzr67 ; https://tinyurl.com/y4pzcwc9).
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
