BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Los_Angeles
X-LIC-LOCATION:America/Los_Angeles
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20240626T180033Z
LOCATION:2010\, 2nd Floor
DTSTART;TZID=America/Los_Angeles:20240626T142400
DTEND;TZID=America/Los_Angeles:20240626T144200
UID:dac_DAC 2024_sess189_ESS017@linklings.com
SUMMARY:Automated Generation of SSD Stress Tests Using Offline Reinforceme
 nt Learning
DESCRIPTION:Embedded Systems and Software\n\nSunghee Lee (Samsung)\n\nRein
 forcement learning has demonstrated optimization performance in various si
 mulation environments, yet there has been limited evidence of its effectiv
 eness in real-world scenarios.\n\nIn this study, we applied offline reinfo
 rcement learning in an SSD simulator with real product-level complexity. A
 ttempting to design test cases that impose high loads on the SSD, we confi
 rmed a reduction of over 50% in test input quantity compared to random tes
 ting.\n\nTo overcome the high complexity, we transformed the extensive inp
 ut range supported by the product into an optimal range, reflecting produc
 t characteristics. We effectively represented internal information using a
  Graph Neural Network.\n\nWe propose an automated test generation framewor
 k that applies the reuse of trajactiories generated during the agent train
 ing process for training.\n\nTopic: AI, Embedded Systems, Engineering Trac
 ks\n\nSession Chair: Frank Schirrmeister (Synopsys)
END:VEVENT
END:VCALENDAR