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:20240626T180034Z
LOCATION:3008\, 3rd Floor
DTSTART;TZID=America/Los_Angeles:20240625T170000
DTEND;TZID=America/Los_Angeles:20240625T171500
UID:dac_DAC 2024_sess109_RESEARCH1570@linklings.com
SUMMARY:Safe Controller Synthesis for Nonlinear Systems via Reinforcement 
 Learning and PAC Approximation
DESCRIPTION:Research Manuscript\n\nXia Zeng (Southwest University), Banglo
 ng Liu (East China Normal University), Zhenbing Zeng (Shanghai University)
 , Zhiming Liu (Southwest University), and Zhengfeng Yang (East China Norma
 l University)\n\nController synthesis for nonlinear systems is an importan
 t research issue. Deep Neural Network (DNN) control policies obtained thro
 ugh reinforcement learning (RL), though exhibiting good performance in sim
 ulations, cannot be applied to safety-critical systems for lack of formal 
 guarantee. To address this, this paper considers fully utilizing the advan
 tages of RL for complex control tasks to obtain a well-performing DNN cont
 roller. Then, using PAC (Probably Approximately Correct) techniques, a pol
 ynomial surrogate controller with probabilistically controllable approxima
 tion error is obtained. Finally, the safety of the control system under th
 e designed polynomial controller is verified using barrier certificate gen
 eration. Experiments demonstrate the effectiveness of our method in genera
 ting controllers with safety guarantees for systems with high dimensions a
 nd degrees.\n\nTopic: Design\n\nKeyword: Design of Cyber-physical Systems 
 and IoT\n\nSession Chairs: Chen Liu (Intel Corporation) and Jakub Szefer (
 Yale University)
END:VEVENT
END:VCALENDAR
