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:20181221T160726Z
LOCATION:D163
DTSTART;TZID=America/Chicago:20181111T113000
DTEND;TZID=America/Chicago:20181111T120000
UID:submissions.supercomputing.org_SC18_sess159_ws_indis104@linklings.com
SUMMARY:Fast Detection of Elephant Flows with Dirichlet-Categorical Infere
 nce
DESCRIPTION:Workshop\nArchitectures, Networks, Security, Workshop Reg Pass
 \n\nFast Detection of Elephant Flows with Dirichlet-Categorical Inference\
 n\nGudibanda, Ros-Giralt, Commike, Lethin\n\nThe problem of elephant flow 
 detection is a longstanding research area with the goal of quickly identif
 ying flows in a network that are large enough to affect the quality of ser
 vice of smaller flows. Past work in this field has largely been either dom
 ain-specific, based on thresholds for a specific flow size metric, or requ
 ired several hyperparameters, reducing their ease of adaptation to the gre
 at variety of traffic distributions present in real-world networks. In thi
 s paper, we present an approach to elephant flow detection that avoids the
 se limitations, utilizing the rigorous framework of Bayesian inference. By
  observing packets sampled from the network, we use Dirichlet-Categorical 
 inference to calculate a posterior distribution that explicitly captures o
 ur uncertainty about the sizes of each flow. We then use this posterior di
 stribution to find the most likely subset of elephant flows under this pro
 babilistic model. Our algorithm rapidly converges to the optimal sampling 
 rate at a speed O(1/n), where n is the number of packet samples received, 
 and the only hyperparameter required is the targeted detection likelihood,
  defined as the probability of correctly inferring all the elephant flows.
  Compared to the state-of-the-art based on static sampling rate, we show a
  reduction in error rate by a factor of 20 times. The proposed method of D
 irichlet-Categorical inference provides a novel, powerful framework to ele
 phant flow detection that is both highly accurate and probabilistically me
 aningful.
URL:https://sc18.supercomputing.org/presentation/?id=ws_indis104&sess=sess
 159
END:VEVENT
END:VCALENDAR