BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:America/Denver
BEGIN:DAYLIGHT
DTSTART:20260308T030000
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3
TZNAME:MDT
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20251102T010000
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11
TZNAME:MST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20251124T233232Z
UID:299FC94E-0E82-4AF9-AEE0-B9884A60FC91
DTSTART;TZID=America/Denver:20251121T110000
DTEND;TZID=America/Denver:20251121T120000
DESCRIPTION:Recent research shows large-scale AI-centric data centers could
  experience rapid fluctuations in power demand due to varying computation 
 loads\, such as sudden spikes from inference or interruption of training l
 arge language models (LLMs). As a consequence\, such huge and fluctuating 
 power demand pose significant challenges to both data center and power uti
 lity operation. Accurate short-term power forecasting allows data centers 
 and utilities to dynamically allocate resources and power large computing 
 clusters as required. However\, due to the complex data center power usage
  patterns and the black-box nature of the underlying AI algorithms running
  in data centers\, explicit modeling of AI-data center is quite challengin
 g.\n\nAlternatively\, to deal with this emerging load forecasting problem\
 , we propose a data-driven workflow to model and predict the short-term el
 ectricity load in an AI-data center\, and such workflow is compatible with
  learning-based algorithms such as LSTM\, GRU\, 1D-CNN. We validate our fr
 amework\, which achieves decent accuracy on data center GPU short-term pow
 er consumption. This provides opportunity for improved power management an
 d sustainable data center operations.\n\nCo-sponsored by: Resilience and C
 lean Energy Systems (RCES)\n\nSpeaker(s): Mariam Mughees\n\nVirtual: https
 ://events.vtools.ieee.org/m/513703
LOCATION:Virtual: https://events.vtools.ieee.org/m/513703
ORGANIZER:xiaotin5@ualberta.ca
SEQUENCE:32
SUMMARY:Short-Term Load Forecasting for AI-Data Center
URL;VALUE=URI:https://events.vtools.ieee.org/m/513703
X-ALT-DESC:Description: <br /><p class="MsoNormal" style="text-align: justi
 fy\; text-justify: inter-ideograph\;">Recent research shows large-scale AI
 -centric data centers could experience rapid fluctuations in power demand 
 due to varying computation loads\, such as sudden spikes from inference or
  interruption of training large language models (LLMs). As a consequence\,
  such huge and fluctuating power demand pose significant challenges to bot
 h data center and power utility operation. Accurate short-term power forec
 asting allows data centers and utilities to dynamically allocate resources
  and power large computing clusters as required. However\, due to the comp
 lex data center power usage patterns and the black-box nature of the under
 lying AI algorithms running in data centers\, explicit modeling of AI-data
  center is quite challenging.&nbsp\;&nbsp\;</p>\n<p class="MsoNormal" styl
 e="text-align: justify\; text-justify: inter-ideograph\;">Alternatively\, 
 to deal with this emerging load forecasting problem\, we propose a data-dr
 iven workflow to model and predict the short-term electricity load in an A
 I-data center\, and such workflow is compatible with learning-based algori
 thms such as LSTM\, GRU\,&nbsp\;1D-CNN. We validate our framework\, which 
 achieves decent accuracy on data center GPU short-term power consumption. 
 This provides opportunity for improved power management and sustainable da
 ta center operations.</p>
END:VEVENT
END:VCALENDAR