BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Domain Adaptation in NLP - Towards Adaptation to Any Domain - Eyal
  Ben-David\, Technion - Israel Institute of Technology
DTSTART:20210429T100000Z
DTEND:20210429T110000Z
UID:TALK159814@talks.cam.ac.uk
CONTACT:Marinela Parovic
DESCRIPTION:Natural Language Processing algorithms have made incredible pr
 ogress\, but they still struggle when applied to out-of-distribution examp
 les. In this work\, we address a challenging and underexplored version of 
 this domain adaptation problem\, where an algorithm is trained on several 
 source domains\, and then applied to examples from an unseen domain that i
 s unknown at training time. Particularly\, no examples\, labeled or unlabe
 led\, or any other knowledge about the target domain are available to the 
 algorithm at training time. We present PADA: A Prompt-based Autoregressive
  Domain Adaptation algorithm\, based on the T5 model. Given a test example
 \, PADA first generates a unique prompt and then\, conditioned on this pro
 mpt\, labels the example with respect to the NLP task. The prompt is a seq
 uence of unrestricted length\, consisting of pre-defined Domain Related Fe
 atures (DRFs) that characterize each of the source domains. Intuitively\, 
 the prompt is a unique signature that maps the test example to the semanti
 c space spanned by the source domains. In experiments with 3 tasks (text c
 lassification and sequence tagging)\, for a total of 14 multi-source adapt
 ation scenarios\, PADA substantially outperforms strong baselines.
LOCATION:https://cam-ac-uk.zoom.us/j/97599459216?pwd=QTRsOWZCOXRTREVnbTJBd
 XVpOXFvdz09
END:VEVENT
END:VCALENDAR