BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Lennard-Jones Centre
SUMMARY:Improving Data Sub-selection for Supervised Tasks 
 with Principal Covariates Regression - Rose K. Cer
 sonsky\, EPFL
DTSTART;TZID=Europe/London:20220314T143000
DTEND;TZID=Europe/London:20220314T150000
UID:TALK170588AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/170588
DESCRIPTION:Data analyses based on linear methods constitute t
 he simplest\, most robust\, and transparent approa
 ches to the automatic processing of large amounts 
 of data for building supervised or unsupervised ma
 chine learning models. Principal covariates regres
 sion (PCovR) is an underappreciated method that in
 terpolates between principal component analysis an
 d linear regression and can be used to convenientl
 y reveal structure-property relations in terms of 
 simple-to-interpret\, low-dimensional maps. We hav
 e recently introduced methods that incorporate PCo
 vR into two popular data selection approaches\, CU
 R and Farthest Point Sampling\, which iteratively 
 identify the most diverse samples and discriminati
 ng features. While our approach is completely gene
 ral\, here we focus on systems relevant to atomist
 ic simulations\, chemistry\, and materials science
  -- fields where feature and sample selection are 
 an increasingly common practice. Our results show 
 that these selection methods identify data subsets
  that out-perform their unsupervised counterparts-
 -which we demonstrate with models of increasing co
 mplexity\, from ridge regression to kernel ridge r
 egression and finally feed-forward neural networks
 .\n\nThis work pulls from:\n\nStructure-Property M
 aps with Kernel Principal Covariates Regression\;\
 nBA Helfrecht\, RK Cersonsky\, G Fraux\, M Ceriott
 i Machine Learning: Science and Technology 1\n\nIm
 proving Sample and Feature Selection with Principa
 l Covariates Regression\;\nRK Cersonsky\, BA Helfr
 echt\, EA Engel\, S Kliavinek\, M Ceriotti Machine
  Learning: Science and Technology 2
LOCATION:Venue to be confirmed
CONTACT:Dr M. Simoncelli
END:VEVENT
END:VCALENDAR