BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20210402T160105Z
LOCATION:Track 4
DTSTART;TZID=America/New_York:20201119T133000
DTEND;TZID=America/New_York:20201119T140000
UID:submissions.supercomputing.org_SC20_sess178_pap601@linklings.com
SUMMARY:BiQGEMM: Matrix Multiplication with Lookup Table For Binary-Coding
 -Based Quantized DNNs
DESCRIPTION:Paper\n\nBiQGEMM: Matrix Multiplication with Lookup Table For 
 Binary-Coding-Based Quantized DNNs\n\nJeon, Park, Kwon, Kim, Yun...\n\nThe
  number of parameters in deep neural networks (DNNs) is rapidly increasing
  to support complicated tasks and to improve model accuracy. Corresponding
 ly, the amount of computations and required memory footprint increase as w
 ell. Quantization is an efficient method to address such concerns. Unfortu
 nately, commercial processors do not fully support quantization because on
 ly fixed data transfers (such as 32 bits) are allowed. Success of quantiza
 tion in practice, hence, relies on an efficient computation engine design,
  especially for matrix multiplication. In this paper, we propose a novel m
 atrix multiplication method, called BiQGEMM, dedicated to quantized DNNs. 
 BiQGEMM can access multiple quantized weights simultaneously in one instru
 ction. In addition, BiQGEMM pre-computes intermediate results that are hig
 hly redundant when quantization leads to limited available computation spa
 ce. Our extensive experimental results show that BiQGEMM presents higher p
 erformance than conventional schemes when DNNs are quantized.\n\nTag: Data
  Analytics, Compression, and Management, Linear Algebra, Machine Learning,
  Deep Learning and Artificial Intelligence\n\nRegistration Category: Tech 
 Program Reg Pass
END:VEVENT
END:VCALENDAR