BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.bham.ac.uk//v3//EN
BEGIN:VEVENT
CATEGORIES:Optimisation and Numerical Analysis Seminars
SUMMARY:Newton-step-based Hard Thresholding Algorithms - N
an Meng (University of Birmingham)
DTSTART:20191204T120000Z
DTEND:20191204T130000Z
UID:TALK3894AT
URL:/talk/index/3894
DESCRIPTION:Compressive sensing (CS) has been applied in the f
ield of image processing\, pattern recognition and
medical applications. A class of algorithms\, kno
wn as hard thresholding algorithms and first propo
sed by Blumensath and Davies\, have been investiga
ted heavily in CS. The existing hard thresholding
algorithms\, such as the iterative hard thresholdi
ng (IHT) and hard thresholding pursuit (HTP)\, uti
lize the gradient steepest descent direction. In s
olving nonlinear optimization problems\, Newton’s
method is also great of importance. We would consi
der to apply Newton’s method to solve the k-sparse
problem. We apply it by modifying the Hessian wit
h a parameter since the original Hessian is non-in
vertible. In this case\, we establish sufficient c
onditions for the convergence of the NSIHT algorit
hm and NSHTP algorithm. Afterwards\, we investigat
e the numerical performance of the class of Newton
-step-based hard thresholding algorithms.
LOCATION:Strathcona\, SR5
CONTACT:Sergey Sergeev
END:VEVENT
END:VCALENDAR