BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20220812T074335Z
LOCATION:Singapore Room
DTSTART;TZID=Europe/Stockholm:20220629T120000
DTEND;TZID=Europe/Stockholm:20220629T123000
UID:submissions.pasc-conference.org_PASC22_sess156_msa130@linklings.com
SUMMARY:Sparse Quadratic Approximation for Graph Learning
DESCRIPTION:Minisymposium\n\nSparse Quadratic Approximation for Graph Lear
ning\n\nPasadakis, Bollhöfer, Schenk\n\nIn this talk we address the proble
m of learning large graphs
represented by M-matrices via an $\ell_1$-
regularized
Gaussian maximum-likelihood approach. We build on top of
a
state-of-the-art sparse precision matrix estimation package and
introduce two algorithms that learn M-matrices, that can be
subseq
uently used for the estimation of graph Laplacian
matrices. In the fi
rst one we propose an unconstrained
method that follows a post proces
sing approach in order to
learn an M-matrix, and in the second one we
implement a
constrained approach based on sequential quadratic prog
ramming.
We also demonstrate the effectiveness, accuracy, and perform
ance
of both algorithms. Our numerical examples and
comparative
results with modern open-source packages reveal
that the proposed me
thods can accelerate the learning of
graphs by up to 3 orders of magn
itude, while accurately
retrieving the latent graphical structure of
the data.
Furthermore, we conduct large scale case studies for the
clustering of COVID-19 daily cases and the classification of
image
datasets to highlight the applicability in real-world
scenarios.\n\n
Domain: Computer Science and Applied Mathematics
END:VEVENT
END:VCALENDAR