BEGIN:VCALENDAR
CALSCALE:GREGORIAN
VERSION:2.0
METHOD:PUBLISH
PRODID:-//Drupal iCal API//EN
X-WR-TIMEZONE:America/New_York
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
DTSTART:20070311T020000
TZNAME:EDT
TZOFFSETTO:-0400
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
DTSTART:20071104T020000
TZNAME:EST
TZOFFSETTO:-0500
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
SEQUENCE:1
X-APPLE-TRAVEL-ADVISORY-BEHAVIOR:AUTOMATIC
UID:172906
DTSTAMP:20240612T094905Z
DTSTART;TZID=America/New_York:20240619T130000
DTEND;TZID=America/New_York:20240619T140000
URL;TYPE=URI:https://www.wpi.edu/news/calendar/events/computer-science-depa
 rtment-phd-dissertation-defense-yiyang-zhao-efficient-and-sustainable-neur
 al
SUMMARY:Computer Science Department , PhD Dissertation Defense  Yiyang Zhao
  " Efficient and Sustainable Neural Architecture Search"
DESCRIPTION:\nYiyang Zhao\nPhD Candidate\nWPI – Computer Science Department
 \n\nWednesday, June 19, 2024\nTime: 1:00 p.m. – 2:00 p.m.\nZoom:https://wp
 i.zoom.us/j/9958714387\n\nCommittee Members:\n\nDissertation Advisor: Prof
 . Tian Guo, WPI – Computer ScienceProf. Xiangnan Kong, WPI – Computer Scie
 nceProf. Xiaozhong Liu. WPI – Computer ScienceExternal Committee Member: D
 r. Tieying Zhang – Bytedance\nAbstract:\nArtificial intelligence (AI) now 
 plays an indispensable role not only in the computer science domain but al
 so in people’s everyday lives. AI solutions have greatly outperformed conv
 entional methods in many real-world tasks and problems, such as image clas
 sification, object detection, and image segmentation. However, the design 
 of AI models and systems is still reserved for domain experts, which large
 ly restricts the development and spread of AI. This proposal seeks to desi
 gn an AI pipeline to automate the production line of AI to remove this res
 triction, in an efficient and sustainable way, from three aspects. (1)Mult
 i-Objective Neural Architecture Search (NAS) Algorithm. Previous NAS works
  have mainly focused on designing models with good performance metrics whi
 le neglecting other important factors such as the number of parameters. In
  this thesis, we propose to design an effective multi-objective search alg
 orithm for NAS, allowing it to consider multiple factors during the design
  process. (2) Efficient Network Evaluations.\nThe vanilla NAS approach req
 uires training all the searched/sampled neural architectures from scratch 
 to evaluate their performance, incurring a significant amount of time and 
 computational costs, often hundreds to thousands of GPU days. In order to 
 reduce the search cost, we propose a few-shot NAS approach that leverages 
 multiple super-nets as proxies to accurately estimate the performance of m
 odels in a shorter amount of time. (3) Sustainable Neural Architecture Sea
 rch. Energy cost and carbon emissions are major environmental issues in NA
 S. Prior work reports that a single architecture search by NAS can produce
  as much carbon emissions as five cars’ lifetimes. Despite many efforts to
  reduce energy costs in NAS, the carbon emissions can still vary with ener
 gy generation methods, e.g., solar vs. fuel, and locations. Therefore, we 
 propose an adaptive carbon-aware NAS framework that reduces carbon consump
 tion during the search while maintaining good search performance.\n
END:VEVENT
END:VCALENDAR