BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Accelerating computation of SVM and DNN by binary approximation - 
 Hironobu Fujiyoshi (Chubu University) 
DTSTART:20170908T100000Z
DTEND:20170908T110000Z
UID:TALK79751@talks.cam.ac.uk
CONTACT:54031
DESCRIPTION:Object detection involves classification of a huge number of d
 etection windows obtained by raster scanning of an input image.\nIn this t
 alk\, we introduce binary approximation to accelerate the computation of l
 inear SVM for multi-class classification task.\nSince the proposed method 
 can replace real-valued inner-product with binary inner-product computatio
 ns\, it's about 200 times faster than the conventional SVM classifiers.\nW
 e also show that the proposed approach can be extended to enable fast comp
 utation of existing deep neural network models and decrease model size wit
 hout the need for retraining. 
LOCATION:CBL Room BE-438\, Department of Engineering
END:VEVENT
END:VCALENDAR