BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.2.3.1//EN
TZID:Asia/Kolkata
X-WR-TIMEZONE:Asia/Kolkata
BEGIN:VEVENT
UID:10@cds.iisc.ac.in
DTSTART;TZID=Asia/Kolkata:20230824T123000
DTEND;TZID=Asia/Kolkata:20230824T133000
DTSTAMP:20231102T050810Z
URL:https://cds.iisc.ac.in/events/seminar-cds-102-24th-august-qameleon-mul
 tilingual-qa-with-only-5-examples/
SUMMARY:{Seminar} @ CDS: #102 : 24th August: “QAmeleon: Multilingual QA w
 ith Only 5 Examples”
DESCRIPTION:Department of Computational and Data Sciences\nDepartment Semin
 ar\n\nSPEAKER : Priyanka Agrawal (Google Deepmind)\n\nTITLE : “QAmeleon:
  Multilingual QA with Only 5 Examples“\n\nDate &amp\; Time : August 24\,
  2023\, 12:30 PM\n\nVenue : # 102\, CDS Seminar Hall\n\n==================
 =========================================================================\
 nAbstract\nThe availability of large\, high-quality datasets has been one 
 of the main drivers of recent progress in question answering (QA). Such an
 notated datasets however are difficult and costly to collect\, and rarely 
 exist in languages other than English\, rendering QA technology inaccessib
 le to underrepresented languages. An alternative to building large monolin
 gual training datasets is to leverage pre-trained language models (PLMs) u
 nder a few-shot learning setting. Our approach\, QAmeleon\, uses a PLM to 
 automatically generate multilingual data upon which QA models are trained\
 , thus avoiding costly annotation. Prompt tuning the PLM for data synthesi
 s with only five examples per language delivers accuracy superior to trans
 lation-based baselines\, bridges nearly 60% of the gap between an English-
 only baseline and a fully supervised upper bound trained on almost 50\,000
  hand labeled examples\, and always leads to substantial improvements comp
 ared to fine-tuning a QA model directly on labeled examples in low resourc
 e settings. Experiments on the TyDiQA-GoldP and MLQA benchmarks show that 
 few-shot prompt tuning for data synthesis scales across languages and is a
  viable alternative to large-scale annotation.\nBiography\nPriyanka Agrawa
 l is a Research Scientist at the Google Deepmind in London\, formally part
  of Google Brain\, and is focused on building responsible Generative AI mo
 dels and scaling them to underrepresented languages. Prior to that she was
  a Senior Researcher and Lead at Booking.com and IBM Research Labs\, where
  she was driving work in cross-domain transfer and representation learning
 . She is also an alumni from IISc from the CSA department. Her work is pub
 lished at top tier ML and NLP conferences like NeurIPS\, ACL and she holds
  25+ US Patents. Priyanka also serves as Area Chair and PC member at these
  conferences and has been an invited panelist and speaker at various ML/NL
 P and diversity forums.\n\nHost Faculty: Dr. Danish Pruthi\n==============
 ====================================================================\n\nAL
 L ARE WELCOME
CATEGORIES:Events,Talks
END:VEVENT
BEGIN:VTIMEZONE
TZID:Asia/Kolkata
X-LIC-LOCATION:Asia/Kolkata
BEGIN:STANDARD
DTSTART:20220824T123000
TZOFFSETFROM:+0530
TZOFFSETTO:+0530
TZNAME:IST
END:STANDARD
END:VTIMEZONE
END:VCALENDAR