BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Denver
X-LIC-LOCATION:America/Denver
BEGIN:DAYLIGHT
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
TZNAME:MDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
TZNAME:MST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20200129T163559Z
LOCATION:712
DTSTART;TZID=America/Denver:20191117T154500
DTEND;TZID=America/Denver:20191117T161500
UID:submissions.supercomputing.org_SC19_sess107_ws_cafcw104@linklings.com
SUMMARY:Deep Kernel Learning for Information Extraction from Cancer Pathol
 ogy Reports
DESCRIPTION:Workshop\n\nDeep Kernel Learning for Information Extraction fr
 om Cancer Pathology Reports\n\nAgrawal, Dubey, Tourassi, Hinkle\n\nCancer 
 pathology reports comprise a rich source of data for surveilling cancer in
 cidents and tracking cancer trends across the United States. Cancer regist
 ries manually extract key pieces of information from these reports includi
 ng tumor site, histology, laterality, behavior, grade, and metastatic stat
 us. Automating this task is critical for an efficient and scalable process
 ing pipeline of these reports. Deep neural networks have recently been sho
 wn to perform well on this information extraction task by casting it as a 
 document classification problem. However, neural networks are prone to ove
 rfitting in low-sample regimes and are unable to quantify their own uncert
 ainty. Deep kernel learning (DKL) has recently emerged as a simple and sca
 lable paradigm to hybridize deep neural networks and Bayesian models, whic
 h may help to remedy some of these shortcomings of neural networks. A DKL 
 model is obtained by feeding a neural network feature extractor into a Gau
 ssian process (GP) classifier and training the resulting model with gradie
 nt descent in a variational inference framework. In this project, we build
  a DKL model with a shallow-wide convolutional neural network (CNN) featur
 e extractor and use it to extract primary tumor site information from a da
 taset of de-identified cancer pathology reports. As far as we are aware, t
 his marks the first application of DKL to document classification. Our DKL
  model outperforms the state-of-the-art CNN on this dataset. We also show 
 that pretraining a CNN with the weights of a DKL model boosts performance,
  suggesting that DKL is beneficial not just because of GP inference at tes
 t time but also because DKL is able to extract better feature representati
 ons from the pathology reports through Bayesian training. We conclude that
  DKL has the potential to boost the performance of neural networks for inf
 ormation extraction on pathology reports while requiring little modificati
 on of the original network architecture, and that DKL can offer a path for
 ward to develop scalable deep Bayesian models for such tasks.\n\nTag: Work
 shop Reg Pass, AI, Bioinformatics, Cancer, Computational Biology, Programm
 ing Systems\n\nRegistration Category: Workshop Reg Pass, AI, Bioinformatic
 s, Cancer, Computational Biology, Programming Systems
URL:https://sc19.supercomputing.org/presentation/?id=ws_cafcw104&sess=sess
 107
END:VEVENT
END:VCALENDAR