Publications
Detailed Information
Brick-by-Brick: Combinatorial Construction with Deep Reinforcement Learning
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Chung, Hyunsoo | - |
dc.contributor.author | Kim, Jungtaek | - |
dc.contributor.author | Knyazev, Boris | - |
dc.contributor.author | Lee, Jinhwi | - |
dc.contributor.author | Taylor, Graham W. | - |
dc.contributor.author | Park, Jaesik | - |
dc.contributor.author | Cho, Minsu | - |
dc.date.accessioned | 2024-05-09T04:12:54Z | - |
dc.date.available | 2024-05-09T04:12:54Z | - |
dc.date.created | 2024-05-08 | - |
dc.date.created | 2024-05-08 | - |
dc.date.issued | 2021 | - |
dc.identifier.citation | Advances in Neural Information Processing Systems, Vol.34, pp.5745-5757 | - |
dc.identifier.issn | 1049-5258 | - |
dc.identifier.uri | https://hdl.handle.net/10371/201300 | - |
dc.description.abstract | Discovering 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.language | 영어 | - |
dc.publisher | Neural information processing systems foundation | - |
dc.title | Brick-by-Brick: Combinatorial Construction with Deep Reinforcement Learning | - |
dc.type | Article | - |
dc.citation.journaltitle | Advances in Neural Information Processing Systems | - |
dc.identifier.wosid | 000901616404003 | - |
dc.identifier.scopusid | 2-s2.0-85131738561 | - |
dc.citation.endpage | 5757 | - |
dc.citation.startpage | 5745 | - |
dc.citation.volume | 34 | - |
dc.description.isOpenAccess | N | - |
dc.contributor.affiliatedAuthor | Park, Jaesik | - |
dc.type.docType | Conference Paper | - |
dc.description.journalClass | 1 | - |
- Appears in Collections:
- Files in This Item:
- There are no files associated with this item.
Item View & Download Count
Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.