BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20230124T171525Z
LOCATION:D174
DTSTART;TZID=America/Chicago:20221116T161500
DTEND;TZID=America/Chicago:20221116T163000
UID:submissions.supercomputing.org_SC22_sess227_spostu102@linklings.com
SUMMARY:Predicting Scientific Data Popularity Using dCache Logs
DESCRIPTION:ACM Student Research Competition: Graduate Poster, ACM Student
  Research Competition: Undergraduate Poster, Posters\n\nPredicting Scienti
 fic Data Popularity Using dCache Logs\n\nBellavita\n\nThe dCache installat
 ion is a storage management system that acts as a disk cache for high-ener
 gy physics (HEP) data. Storagespace on dCache is limited relative to persi
 stent storage devices, therefore, a heuristic is needed to determine what 
 data should be kept in the cache. A good cache policy would keep frequentl
 y accessed data in the cache, but this requires knowledge of future datase
 t popularity. We present methods for forecasting the number of times a dat
 aset stored on dCache will be accessed in the future. We present a deep ne
 ural network that can predict future dataset accesses accurately, reportin
 g a final normalized loss of 4.6e-8. We present a set of algorithms that c
 an forecast future dataset accesses given an access sequence. Included are
  two novel algorithms, Backup Predictor and Last N Successors, that outper
 form other file prediction algorithms. Findings suggest that it is possibl
 e to anticipate dataset popularity in advance.\n\nSession Format: Recorded
 \n\nRegistration Category: Tech Program Reg Pass
END:VEVENT
END:VCALENDAR