Show simple item record

dc.contributor.author
Abad, Javier
dc.contributor.author
Bhatt, Umang
dc.contributor.author
Weller, Adrian
dc.contributor.author
Cherubin, Giovanni
dc.contributor.editor
Williams, Brian
dc.contributor.editor
Chen, Yiling
dc.contributor.editor
Neville, Jennifer
dc.date.accessioned
2024-10-01T16:08:19Z
dc.date.available
2023-08-25T06:05:30Z
dc.date.available
2023-09-01T11:38:23Z
dc.date.available
2024-10-01T16:08:19Z
dc.date.issued
2023-06-27
dc.identifier.isbn
978-1-57735-880-0
en_US
dc.identifier.other
10.1609/aaai.v37i6.25814
en_US
dc.identifier.uri
http://hdl.handle.net/20.500.11850/628143
dc.description.abstract
Conformal prediction (CP) is a wrapper around traditional machine learning models, giving coverage guarantees under the sole assumption of exchangeability; in classification problems, a CP guarantees that the error rate is at most a chosen significance level ε, irrespective of whether the underlying model is misspecified. However, the prohibitive computational costs of full CP led researchers to design scalable alternatives, which alas do not attain the same guarantees or statistical power of full CP. In this paper, we use influence functions to efficiently approximate full CP. We prove that our method is a consistent approximation of full CP, and empirically show that the approximation error becomes smaller as the training set increases; e.g., for 1, 000 training points the two methods output p-values that are < 0.001 apart: a negligible error for any practical application. Our methods enable scaling full CP to large real-world datasets. We compare our full CP approximation (ACP) to mainstream CP alternatives, and observe that our method is computationally competitive whilst enjoying the statistical predictive power of full CP.
en_US
dc.language.iso
en
en_US
dc.publisher
AAAI
en_US
dc.title
Approximating Full Conformal Prediction at Scale via Influence Functions
en_US
dc.type
Conference Paper
dc.date.published
2023-06-26
ethz.book.title
Proceedings of the 37th AAAI Conference on Artificial Intelligence
en_US
ethz.journal.volume
37
en_US
ethz.journal.issue
6
en_US
ethz.pages.start
6631
en_US
ethz.pages.end
6639
en_US
ethz.event
AAAI Conference on Artificial Intelligence (AAAI-23)
en_US
ethz.event.location
Washington, DC, USA
en_US
ethz.event.date
February 7-14, 2023
en_US
ethz.identifier.wos
ethz.identifier.scopus
ethz.publication.place
Washington, DC
en_US
ethz.publication.status
published
en_US
ethz.date.deposited
2023-08-25T06:05:32Z
ethz.source
SCOPUS
ethz.eth
yes
en_US
ethz.availability
Metadata only
en_US
ethz.rosetta.installDate
2023-09-01T11:38:24Z
ethz.rosetta.lastUpdated
2023-09-01T11:38:24Z
ethz.rosetta.exportRequired
true
ethz.rosetta.versionExported
true
ethz.COinS
ctx_ver=Z39.88-2004&amp;rft_val_fmt=info:ofi/fmt:kev:mtx:journal&amp;rft.atitle=Approximating%20Full%20Conformal%20Prediction%20at%20Scale%20via%20Influence%20Functions&amp;rft.date=2023-06-27&amp;rft.volume=37&amp;rft.issue=6&amp;rft.spage=6631&amp;rft.epage=6639&amp;rft.au=Abad,%20Javier&amp;Bhatt,%20Umang&amp;Weller,%20Adrian&amp;Cherubin,%20Giovanni&amp;rft.isbn=978-1-57735-880-0&amp;rft.genre=proceeding&amp;rft_id=info:doi/10.1609/aaai.v37i6.25814&amp;rft.btitle=Proceedings%20of%20the%2037th%20AAAI%20Conference%20on%20Artificial%20Intelligence
 Search print copy at ETH Library

Files in this item

FilesSizeFormatOpen in viewer

There are no files associated with this item.

Publication type

Show simple item record