Show simple item record

dc.contributor.authorRei, Mareken
dc.contributor.authorSøgaard, Andersen
dc.date.accessioned2019-01-11T00:30:45Z
dc.date.available2019-01-11T00:30:45Z
dc.identifier.urihttps://www.repository.cam.ac.uk/handle/1810/287795
dc.description.abstractCan attention- or gradient-based visualization techniques be used to infer token-level labels for binary sequence tagging problems, using networks trained only on sentence-level labels? We construct a neural network architecture based on soft attention, train it as a binary sentence classifier and evaluate against token-level annotation on four different datasets. Inferring token labels from a network provides a method for quantitatively evaluating what the model is learning, along with generating useful feedback in assistance systems. Our results indicate that attention-based methods are able to predict token-level labels more accurately, compared to gradient-based methods, sometimes even rivaling the supervised oracle network.
dc.description.sponsorshipERC Nvidia
dc.titleZero-shot Sequence Labeling: Transferring Knowledge from Sentences to Tokensen
dc.typeConference Object
dc.identifier.doi10.17863/CAM.35110
dcterms.dateAccepted2018-02-14en
rioxxterms.versionAM
rioxxterms.licenseref.urihttp://www.rioxx.net/licenses/all-rights-reserveden
rioxxterms.licenseref.startdate2018-02-14en
rioxxterms.typeConference Paper/Proceeding/Abstracten
pubs.funder-project-idCambridge Assessment (unknown)
pubs.conference-nameThe 16th Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologiesen
pubs.conference-start-date2018-06-01en
cam.orpheus.counter7*
rioxxterms.freetoread.startdate2022-01-10


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record