Variational Bayesian Optimal Experimental Design

Variational Bayesian Optimal Experimental Design. Foster, A., Jankowiak, M., Bingham, E., Horsfall, P., Teh, Y. W., Rainforth, T., & Goodman, N. In Advances in Neural Information Processing Systems, volume 32, 2019. Curran Associates, Inc..

Paper abstract bibtex

Bayesian optimal experimental design (BOED) is a principled framework for making efficient use of limited experimental resources. Unfortunately, its applicability is hampered by the difficulty of obtaining accurate estimates of the expected information gain (EIG) of an experiment. To address this, we introduce several classes of fast EIG estimators by building on ideas from amortized variational inference. We show theoretically and empirically that these estimators can provide significant gains in speed and accuracy over previous approaches. We further demonstrate the practicality of our approach on a number of end-to-end experiments.

@inproceedings{foster2019,
	title = {Variational {Bayesian} {Optimal} {Experimental} {Design}},
	volume = {32},
	url = {https://proceedings.neurips.cc/paper_files/paper/2019/hash/d55cbf210f175f4a37916eafe6c04f0d-Abstract.html},
	abstract = {Bayesian optimal experimental design (BOED) is a principled framework for making efficient use of limited experimental resources. Unfortunately, its applicability is hampered by the difficulty of obtaining accurate estimates of the expected information gain (EIG) of an experiment. To address this, we introduce several classes of fast EIG estimators by building on ideas from amortized variational inference. We show theoretically and empirically that these estimators can provide significant gains in speed and accuracy over previous approaches. We further demonstrate the practicality of our approach on a number of end-to-end experiments.},
	urldate = {2024-11-20},
	booktitle = {Advances in {Neural} {Information} {Processing} {Systems}},
	publisher = {Curran Associates, Inc.},
	author = {Foster, Adam and Jankowiak, Martin and Bingham, Elias and Horsfall, Paul and Teh, Yee Whye and Rainforth, Thomas and Goodman, Noah},
	year = {2019},
	file = {Full Text PDF:/Users/lcneuro/Zotero/storage/PS5E9WZH/Foster et al. - 2019 - Variational Bayesian Optimal Experimental Design.pdf:application/pdf},
}

Downloads: 0

{"_id":"6DeRvpngNom9zpCMy","bibbaseid":"foster-jankowiak-bingham-horsfall-teh-rainforth-goodman-variationalbayesianoptimalexperimentaldesign-2019","authorIDs":[],"author_short":["Foster, A.","Jankowiak, M.","Bingham, E.","Horsfall, P.","Teh, Y. W.","Rainforth, T.","Goodman, N."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","title":"Variational Bayesian Optimal Experimental Design","volume":"32","url":"https://proceedings.neurips.cc/paper_files/paper/2019/hash/d55cbf210f175f4a37916eafe6c04f0d-Abstract.html","abstract":"Bayesian optimal experimental design (BOED) is a principled framework for making efficient use of limited experimental resources. Unfortunately, its applicability is hampered by the difficulty of obtaining accurate estimates of the expected information gain (EIG) of an experiment. To address this, we introduce several classes of fast EIG estimators by building on ideas from amortized variational inference. We show theoretically and empirically that these estimators can provide significant gains in speed and accuracy over previous approaches. We further demonstrate the practicality of our approach on a number of end-to-end experiments.","urldate":"2024-11-20","booktitle":"Advances in Neural Information Processing Systems","publisher":"Curran Associates, Inc.","author":[{"propositions":[],"lastnames":["Foster"],"firstnames":["Adam"],"suffixes":[]},{"propositions":[],"lastnames":["Jankowiak"],"firstnames":["Martin"],"suffixes":[]},{"propositions":[],"lastnames":["Bingham"],"firstnames":["Elias"],"suffixes":[]},{"propositions":[],"lastnames":["Horsfall"],"firstnames":["Paul"],"suffixes":[]},{"propositions":[],"lastnames":["Teh"],"firstnames":["Yee","Whye"],"suffixes":[]},{"propositions":[],"lastnames":["Rainforth"],"firstnames":["Thomas"],"suffixes":[]},{"propositions":[],"lastnames":["Goodman"],"firstnames":["Noah"],"suffixes":[]}],"year":"2019","file":"Full Text PDF:/Users/lcneuro/Zotero/storage/PS5E9WZH/Foster et al. - 2019 - Variational Bayesian Optimal Experimental Design.pdf:application/pdf","bibtex":"@inproceedings{foster2019,\n\ttitle = {Variational {Bayesian} {Optimal} {Experimental} {Design}},\n\tvolume = {32},\n\turl = {https://proceedings.neurips.cc/paper_files/paper/2019/hash/d55cbf210f175f4a37916eafe6c04f0d-Abstract.html},\n\tabstract = {Bayesian optimal experimental design (BOED) is a principled framework for making efficient use of limited experimental resources. Unfortunately, its applicability is hampered by the difficulty of obtaining accurate estimates of the expected information gain (EIG) of an experiment. To address this, we introduce several classes of fast EIG estimators by building on ideas from amortized variational inference. We show theoretically and empirically that these estimators can provide significant gains in speed and accuracy over previous approaches. We further demonstrate the practicality of our approach on a number of end-to-end experiments.},\n\turldate = {2024-11-20},\n\tbooktitle = {Advances in {Neural} {Information} {Processing} {Systems}},\n\tpublisher = {Curran Associates, Inc.},\n\tauthor = {Foster, Adam and Jankowiak, Martin and Bingham, Elias and Horsfall, Paul and Teh, Yee Whye and Rainforth, Thomas and Goodman, Noah},\n\tyear = {2019},\n\tfile = {Full Text PDF:/Users/lcneuro/Zotero/storage/PS5E9WZH/Foster et al. - 2019 - Variational Bayesian Optimal Experimental Design.pdf:application/pdf},\n}\n\n","author_short":["Foster, A.","Jankowiak, M.","Bingham, E.","Horsfall, P.","Teh, Y. W.","Rainforth, T.","Goodman, N."],"bibbaseid":"foster-jankowiak-bingham-horsfall-teh-rainforth-goodman-variationalbayesianoptimalexperimentaldesign-2019","role":"author","urls":{"Paper":"https://proceedings.neurips.cc/paper_files/paper/2019/hash/d55cbf210f175f4a37916eafe6c04f0d-Abstract.html"},"metadata":{"authorlinks":{}},"downloads":0},"bibtype":"inproceedings","biburl":"https://bibbase.org/f/8yBxW5neHauDryu9w/LCNeuro Site.bib","creationDate":"2020-01-27T02:13:33.790Z","downloads":0,"keywords":[],"search_terms":["variational","bayesian","optimal","experimental","design","foster","jankowiak","bingham","horsfall","teh","rainforth","goodman"],"title":"Variational Bayesian Optimal Experimental Design","year":2019,"dataSources":["hEoKh4ygEAWbAZ5iy","Lou3KcoFGgAb5MdCt","ZHHpjMFTPYP8NqJif"]}