BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Australia/Melbourne
X-LIC-LOCATION:Australia/Melbourne
BEGIN:DAYLIGHT
TZOFFSETFROM:+1000
TZOFFSETTO:+1100
TZNAME:AEDT
DTSTART:19721003T020000
RRULE:FREQ=YEARLY;BYMONTH=4;BYDAY=1SU
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:19721003T020000
TZOFFSETFROM:+1100
TZOFFSETTO:+1000
TZNAME:AEST
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20240214T070241Z
LOCATION:Darling Harbour Theatre\, Level 2 (Convention Centre)
DTSTART;TZID=Australia/Melbourne:20231212T093000
DTEND;TZID=Australia/Melbourne:20231212T124500
UID:siggraphasia_SIGGRAPH Asia 2023_sess209_papers_819@linklings.com
SUMMARY:Learning Gradient Fields for Scalable and Generalizable Irregular 
 Packing
DESCRIPTION:Technical Papers\n\nTianyang Xue (Shandong University), Mingdo
 ng Wu (Peking University), Lin Lu and Haoxuan Wang (Shandong University), 
 and Hao Dong and Baoquan Chen (Peking University)\n\nThe packing problem, 
 also known as cutting or nesting, has diverse applications in logistics, m
 anufacturing, layout design, and atlas generation. It involves arranging i
 rregularly shaped pieces to minimize waste while avoiding overlap. Recent 
 advances in machine learning, particularly reinforcement learning, have sh
 own promise in addressing the packing problem. In this work, we delve deep
 er into a novel machine learning-based approach that formulates the packin
 g problem as conditional generative modeling. To tackle the challenges of 
 irregular packing, including object validity constraints and collision avo
 idance, our method employs the score-based diffusion model to learn a seri
 es of gradient fields. These gradient fields encode the correlations betwe
 en constraint satisfaction and the spatial relationships of polygons, lear
 ned from teacher examples. During the testing phase, packing solutions are
  generated using a coarse-to-fine refinement mechanism guided by the learn
 ed gradient fields. To enhance packing feasibility and optimality, we intr
 oduce two key architectural designs: multi-scale feature extraction and co
 arse-to-fine relation extraction. We conduct experiments on two typical in
 dustrial packing domains, considering translations only. Empirically, our 
 approach demonstrates spatial utilization rates comparable to, or even sur
 passing, those achieved by the teacher algorithm responsible for training 
 data generation. Additionally, it exhibits some level of generalization to
  shape variations. We are hopeful that this method could pave the way for 
 new possibilities in solving the packing problem.\n\nRegistration Category
 : Full Access, Enhanced Access, Trade Exhibitor, Experience Hall Exhibitor
URL:https://asia.siggraph.org/2023/full-program?id=papers_819&sess=sess209
END:VEVENT
END:VCALENDAR