BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:US/Mountain
BEGIN:DAYLIGHT
DTSTART:20210314T030000
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3
TZNAME:MDT
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20211107T010000
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11
TZNAME:MST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20211015T190636Z
UID:1E569F30-F98D-40B8-8AB4-61E9A9481D60
DTSTART;TZID=US/Mountain:20211014T190000
DTEND;TZID=US/Mountain:20211014T200000
DESCRIPTION:Artificial intelligence (AI) is ubiquitous (self-driving cars\,
  smart appliances\, health monitoring). Estimates by OpenAI predict explos
 ive growth of computational requirements associated with AI by a factor of
  100× every two years\, which is a 50×faster rate than Moore’s law gov
 erning the evolution of the chip industry. As AI becomes increasingly reli
 ant on deep learning neural networks (DNN)\, energy efficient hardware ass
 umes a position of paramount importance. Present day AI tends to dissipate
  an enormous amount of energy for training and inference (300 Google searc
 hes consume enough energy to boil 1 liter of water at room temperature). A
 gainst this backdrop\, there is a serious desire to identify a technology 
 that can reduce energy consumption dramatically in DNNs.\n\nA promising ca
 ndidate for such a technology is “straintronics” which relies on the m
 anipulation of magnetic states in magnetostrictive nanomagnets via electri
 cally-generated strain to elicit myriad non-Boolean computing activities\,
  such as in DNN. The energy-delay product associated with switching a nano
 magnet’s magnetic state using strain is ~10^-27 J-s at room temperature\
 , which is one order of magnitude lower than that associated with switchin
 g a modern day FINFET\, and more than three orders of magnitude lower than
  that associated with switching magnetization with spin-orbit torques or s
 pin transfer torques in STT-RAM.\n\nOur collaborators and we have develope
 d many constructs for processing and communicating information with strain
 tronics for the purpose of AI. They include neurons and synapses dissipati
 ng miniscule amount of energy\, compact restricted Boltzmann machines for 
 image classification\, ternary content addressable memory with drastically
  reduced footprint\, hardware accelerators for image processing\, Bayesian
  inference engines\, correlators/anti-correlators for probabilistic bits\,
  bit comparators for cyber-security applications\, analog computing\, and 
 (non-volatile) matrix multipliers for machine learning. This talk will des
 cribe some of these advances.\n\nBoise\, Idaho\, United States\, Virtual: 
 https://events.vtools.ieee.org/m/284139
LOCATION:Boise\, Idaho\, United States\, Virtual: https://events.vtools.iee
 e.org/m/284139
ORGANIZER:pi-boson@ieee.org
SEQUENCE:2
SUMMARY:Straintronics: Manipulating nanomagnets with strain for causal inte
 lligence
URL;VALUE=URI:https://events.vtools.ieee.org/m/284139
X-ALT-DESC:Description: <br /><p>Artificial intelligence (AI) is ubiquitous
  (self-driving cars\, smart appliances\, health monitoring). Estimates by 
 OpenAI predict explosive growth of computational requirements associated w
 ith AI by a factor of 100&times\; every two years\, which is a 50&times\;f
 aster rate than Moore&rsquo\;s law governing the evolution of the chip ind
 ustry. As AI becomes increasingly reliant on deep learning neural networks
  (DNN)\, energy efficient hardware assumes a position of paramount importa
 nce. Present day AI tends to dissipate an enormous amount of energy for tr
 aining and inference (300 Google searches consume enough energy to boil 1 
 liter of water at room temperature). Against this backdrop\, there is a se
 rious desire to identify a technology that can reduce energy consumption d
 ramatically in DNNs.</p>\n<p>A promising candidate for such a technology i
 s &ldquo\;straintronics&rdquo\; which relies on the manipulation of magnet
 ic states in magnetostrictive nanomagnets via electrically-generated strai
 n to elicit myriad non-Boolean computing activities\, such as in DNN. The 
 energy-delay product associated with switching a nanomagnet&rsquo\;s magne
 tic state using strain is ~10^-27 J-s at room temperature\, which is one o
 rder of magnitude lower than that associated with switching a modern day F
 INFET\, and more than three orders of magnitude lower than that associated
  with switching magnetization with spin-orbit torques or spin transfer tor
 ques in STT-RAM.</p>\n<p>Our collaborators and we have developed many cons
 tructs for processing and communicating information with straintronics for
  the purpose of AI. They include neurons and synapses dissipating miniscul
 e amount of energy\, compact restricted Boltzmann machines for image class
 ification\, ternary content addressable memory with drastically reduced fo
 otprint\, hardware accelerators for image processing\, Bayesian inference 
 engines\, correlators/anti-correlators for probabilistic bits\, bit compar
 ators for cyber-security applications\, analog computing\, and (non-volati
 le) matrix multipliers for machine learning. This talk will describe some 
 of these advances.</p>\n<p>&nbsp\;</p>
END:VEVENT
END:VCALENDAR