BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Asia/Hong_Kong
X-LIC-LOCATION:Asia/Hong_Kong
BEGIN:STANDARD
TZOFFSETFROM:+0800
TZOFFSETTO:+0800
TZNAME:HKT
DTSTART:19911015T033000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20251218T030653Z
LOCATION:Meeting Room S221\, Level 2
DTSTART;TZID=Asia/Hong_Kong:20251218T104000
DTEND;TZID=Asia/Hong_Kong:20251218T105000
UID:siggraphasia_SIGGRAPH Asia 2025_sess153_papers_1036@linklings.com
SUMMARY:Physics-Based Motion Imitation with Adversarial Differential Discr
 iminators
DESCRIPTION:Ziyu Zhang (Simon Fraser University); Sergey Bashkirov (Sony P
 laystation); Dun Yang and Yi Shi (Simon Fraser University); Michael Taylor
  (Sony Playstation); and Xue Bin Peng (Simon Fraser University, NVIDIA)\n\
 nMulti-objective optimization problems, which require the simultaneous opt
 imization of multiple objectives, are prevalent across numerous applicatio
 ns. Existing multi-objective optimization methods often rely on manually-t
 uned aggregation functions to formulate a joint optimization objective. Th
 e performance of such hand-tuned methods is heavily dependent on careful w
 eight selection, a time-consuming and laborious process. These limitations
  also arise in the setting of reinforcement-learning-based motion tracking
  methods for physically simulated characters, where intricately crafted re
 ward functions are typically used to achieve high-fidelity results. Such s
 olutions not only require domain expertise and significant manual tuning, 
 but also limit the applicability of the resulting reward function across d
 iverse skills. To bridge this gap, we present a novel adversarial multi-ob
 jective optimization technique that is broadly applicable to a range of mu
 lti-objective reinforcement-learning tasks, including motion tracking. Our
  proposed Adversarial Differential Discriminator (ADD) receives a single p
 ositive sample, yet is still effective at guiding the optimization process
 . We demonstrate that our technique can enable characters to closely repli
 cate a variety of acrobatic and agile behaviors, achieving comparable qual
 ity to state-of-the-art motion-tracking methods, without relying on manual
 ly-designed reward functions.\n\nRegistration Category: Full Access, Full 
 Access Supporter\n\nSession Chair: Kai Wang (Simon Fraser University)\n\n
END:VEVENT
END:VCALENDAR