Publications

Detailed Information

Brick-by-Brick: Combinatorial Construction with Deep Reinforcement Learning

DC Field Value Language
dc.contributor.authorChung, Hyunsoo-
dc.contributor.authorKim, Jungtaek-
dc.contributor.authorKnyazev, Boris-
dc.contributor.authorLee, Jinhwi-
dc.contributor.authorTaylor, Graham W.-
dc.contributor.authorPark, Jaesik-
dc.contributor.authorCho, Minsu-
dc.date.accessioned2024-05-09T04:12:54Z-
dc.date.available2024-05-09T04:12:54Z-
dc.date.created2024-05-08-
dc.date.created2024-05-08-
dc.date.issued2021-
dc.identifier.citationAdvances in Neural Information Processing Systems, Vol.34, pp.5745-5757-
dc.identifier.issn1049-5258-
dc.identifier.urihttps://hdl.handle.net/10371/201300-
dc.description.abstractDiscovering a solution in a combinatorial space is prevalent in many real-world problems but it is also challenging due to diverse complex constraints and the vast number of possible combinations. To address such a problem, we introduce a novel formulation, combinatorial construction, which requires a building agent to assemble unit primitives (i.e., LEGO bricks) sequentially - every connection between two bricks must follow a fixed rule, while no bricks mutually overlap. To construct a target object, we provide incomplete knowledge about the desired target (i.e., 2D images) instead of exact and explicit volumetric information to the agent. This problem requires a comprehensive understanding of partial information and long-term planning to append a brick sequentially, which leads us to employ reinforcement learning. The approach has to consider a variable-sized action space where a large number of invalid actions, which would cause overlap between bricks, exist. To resolve these issues, our model, dubbed Brick-by-Brick, adopts an action validity prediction network that efficiently filters invalid actions for an actor-critic network. We demonstrate that the proposed method successfully learns to construct an unseen object conditioned on a single image or multiple views of a target object.-
dc.language영어-
dc.publisherNeural information processing systems foundation-
dc.titleBrick-by-Brick: Combinatorial Construction with Deep Reinforcement Learning-
dc.typeArticle-
dc.citation.journaltitleAdvances in Neural Information Processing Systems-
dc.identifier.wosid000901616404003-
dc.identifier.scopusid2-s2.0-85131738561-
dc.citation.endpage5757-
dc.citation.startpage5745-
dc.citation.volume34-
dc.description.isOpenAccessN-
dc.contributor.affiliatedAuthorPark, Jaesik-
dc.type.docTypeConference Paper-
dc.description.journalClass1-
Appears in Collections:
Files in This Item:
There are no files associated with this item.

Related Researcher

  • College of Engineering
  • Dept. of Computer Science and Engineering
Research Area Computer Graphics, Computer Vision, Machine Learning, Robotics

Altmetrics

Item View & Download Count

  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Share