BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Drupal iCal API//EN
X-WR-CALNAME:Events items teaser
X-WR-TIMEZONE:America/Toronto
BEGIN:VTIMEZONE
TZID:America/Toronto
X-LIC-LOCATION:America/Toronto
BEGIN:DAYLIGHT
TZNAME:EDT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
DTSTART:20190310T070000
END:DAYLIGHT
BEGIN:STANDARD
TZNAME:EST
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
DTSTART:20191103T060000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
UID:69d97fdfe4fce
DTSTART;TZID=America/Toronto:20200205T100000
SEQUENCE:0
TRANSP:TRANSPARENT
DTEND;TZID=America/Toronto:20200205T100000
URL:https://uwaterloo.ca/statistics-and-actuarial-science/events/department
 -seminar-david-kepplinger-university-british
LOCATION:M3 - Mathematics 3 200 University Avenue West Room 3127 Waterloo O
 N N2L 3G1 Canada
SUMMARY:Department seminar by David Kepplinger\, University of British Colu
 mbia
CLASS:PUBLIC
DESCRIPTION:DETECTING THE SIGNAL AMONG NOISE AND CONTAMINATION IN HIGH DIME
 NSIONS\n\nImprovements in biomedical technology and a surge in other data-
 driven\nsciences lead to the collection of increasingly large amounts of d
 ata.\nIn this affluence of data\, contamination is ubiquitous but often\nn
 eglected\, creating substantial risk of spurious scientific\ndiscoveries. 
 Especially in applications with high-dimensional data\,\nfor instance prot
 eomic biomarker discovery\, the impact of\ncontamination on methods for va
 riable selection and estimation can be\nprofound yet difficult to diagnose
 .\n\nIn this talk I present a method for variable selection and estimation
 \nin high-dimensional linear regression models\, leveraging the\nelastic-n
 et penalty for complex data structures. The method is capable\nof harnessi
 ng the collected information even in the presence of\narbitrary contaminat
 ion in the response and the predictors. I showcase\nthe method’s theoret
 ical and practical advantages\, specifically in\napplications with heavy-t
 ailed errors and limited control over the\ndata. I outline efficient algor
 ithms to tackle computational\nchallenges posed by inherently non-convex o
 bjective functions of\nrobust estimators and practical strategies for hype
 r-parameter\nselection\, ensuring scalability of the method and applicabil
 ity to a\nwide range of problems.
DTSTAMP:20260410T225527Z
END:VEVENT
END:VCALENDAR