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:3001\, 3rd Floor
DTSTART;TZID=America/Los_Angeles:20240625T114500
DTEND;TZID=America/Los_Angeles:20240625T120000
UID:dac_DAC 2024_sess103_RESEARCH504@linklings.com
SUMMARY:Go Beyond Black-box Policies: Rethinking the Design of Learning Ag
 ent for Interpretable and Verifiable HVAC Control
DESCRIPTION:Research Manuscript\n\nZhiyu An, Xianzhong Ding, and Wan Du (U
 niversity of California, Merced)\n\nRecent research has shown the potentia
 l of Model-based Reinforcement Learning (MBRL) to enhance energy efficienc
 y of Heating, Ventilation, and Air Conditioning (HVAC) systems. However, e
 xisting methods rely on black-box thermal dynamics models and stochastic o
 ptimizers, lacking reliability guarantees and posing risks to occupant hea
 lth. We address this by redesigning HVAC controllers using decision trees 
 extracted from thermal models and historical data, providing deterministic
 , verifiable, and interpretable policies. Extensive experiments show that 
 our method saves 68.4% more energy and increases human comfort gain by 14.
 8% compared to the state-of-the-art method, plus a 1127x reduction in comp
 utation overhead. Code: https://github.com/30363/Veri-HVAC.\n\nTopic: AI\n
 \nKeyword: AI/ML Algorithms\n\nSession Chairs: Hongyang Jia (Tsinghua Univ
 ersity) and Grace Li Zhang (Technische Universität Darmstadt)
END:VEVENT
END:VCALENDAR