Matches in SemOpenAlex for { <https://semopenalex.org/work/W4287748764> ?p ?o ?g. }
Showing items 1 to 85 of
85
with 100 items per page.
- W4287748764 abstract "Robust perception relies on both bottom-up and top-down signals. Bottom-up signals consist of what's directly observed through sensation. Top-down signals consist of beliefs and expectations based on past experience and short-term memory, such as how the phrase `peanut butter and~...' will be completed. The optimal combination of bottom-up and top-down information remains an open question, but the manner of combination must be dynamic and both context and task dependent. To effectively utilize the wealth of potential top-down information available, and to prevent the cacophony of intermixed signals in a bidirectional architecture, mechanisms are needed to restrict information flow. We explore deep recurrent neural net architectures in which bottom-up and top-down signals are dynamically combined using attention. Modularity of the architecture further restricts the sharing and communication of information. Together, attention and modularity direct information flow, which leads to reliable performance improvements in perceptual and language tasks, and in particular improves robustness to distractions and noisy data. We demonstrate on a variety of benchmarks in language modeling, sequential image classification, video prediction and reinforcement learning that the emph{bidirectional} information flow can improve results over strong baselines." @default.
- W4287748764 created "2022-07-26" @default.
- W4287748764 creator A5013102998 @default.
- W4287748764 creator A5016769717 @default.
- W4287748764 creator A5020059055 @default.
- W4287748764 creator A5030891292 @default.
- W4287748764 creator A5043037494 @default.
- W4287748764 creator A5047726287 @default.
- W4287748764 creator A5072322524 @default.
- W4287748764 creator A5086198262 @default.
- W4287748764 date "2020-06-30" @default.
- W4287748764 modified "2023-09-26" @default.
- W4287748764 title "Learning to Combine Top-Down and Bottom-Up Signals in Recurrent Neural Networks with Attention over Modules" @default.
- W4287748764 doi "https://doi.org/10.48550/arxiv.2006.16981" @default.
- W4287748764 hasPublicationYear "2020" @default.
- W4287748764 type Work @default.
- W4287748764 citedByCount "0" @default.
- W4287748764 crossrefType "posted-content" @default.
- W4287748764 hasAuthorship W4287748764A5013102998 @default.
- W4287748764 hasAuthorship W4287748764A5016769717 @default.
- W4287748764 hasAuthorship W4287748764A5020059055 @default.
- W4287748764 hasAuthorship W4287748764A5030891292 @default.
- W4287748764 hasAuthorship W4287748764A5043037494 @default.
- W4287748764 hasAuthorship W4287748764A5047726287 @default.
- W4287748764 hasAuthorship W4287748764A5072322524 @default.
- W4287748764 hasAuthorship W4287748764A5086198262 @default.
- W4287748764 hasBestOaLocation W42877487641 @default.
- W4287748764 hasConcept C104317684 @default.
- W4287748764 hasConcept C115903868 @default.
- W4287748764 hasConcept C135798126 @default.
- W4287748764 hasConcept C138885662 @default.
- W4287748764 hasConcept C147168706 @default.
- W4287748764 hasConcept C151730666 @default.
- W4287748764 hasConcept C154945302 @default.
- W4287748764 hasConcept C169760540 @default.
- W4287748764 hasConcept C185592680 @default.
- W4287748764 hasConcept C26760741 @default.
- W4287748764 hasConcept C2779136372 @default.
- W4287748764 hasConcept C2779343474 @default.
- W4287748764 hasConcept C2779478453 @default.
- W4287748764 hasConcept C41008148 @default.
- W4287748764 hasConcept C41895202 @default.
- W4287748764 hasConcept C50644808 @default.
- W4287748764 hasConcept C54355233 @default.
- W4287748764 hasConcept C55493867 @default.
- W4287748764 hasConcept C63479239 @default.
- W4287748764 hasConcept C86803240 @default.
- W4287748764 hasConcept C97541855 @default.
- W4287748764 hasConceptScore W4287748764C104317684 @default.
- W4287748764 hasConceptScore W4287748764C115903868 @default.
- W4287748764 hasConceptScore W4287748764C135798126 @default.
- W4287748764 hasConceptScore W4287748764C138885662 @default.
- W4287748764 hasConceptScore W4287748764C147168706 @default.
- W4287748764 hasConceptScore W4287748764C151730666 @default.
- W4287748764 hasConceptScore W4287748764C154945302 @default.
- W4287748764 hasConceptScore W4287748764C169760540 @default.
- W4287748764 hasConceptScore W4287748764C185592680 @default.
- W4287748764 hasConceptScore W4287748764C26760741 @default.
- W4287748764 hasConceptScore W4287748764C2779136372 @default.
- W4287748764 hasConceptScore W4287748764C2779343474 @default.
- W4287748764 hasConceptScore W4287748764C2779478453 @default.
- W4287748764 hasConceptScore W4287748764C41008148 @default.
- W4287748764 hasConceptScore W4287748764C41895202 @default.
- W4287748764 hasConceptScore W4287748764C50644808 @default.
- W4287748764 hasConceptScore W4287748764C54355233 @default.
- W4287748764 hasConceptScore W4287748764C55493867 @default.
- W4287748764 hasConceptScore W4287748764C63479239 @default.
- W4287748764 hasConceptScore W4287748764C86803240 @default.
- W4287748764 hasConceptScore W4287748764C97541855 @default.
- W4287748764 hasLocation W42877487641 @default.
- W4287748764 hasOpenAccess W4287748764 @default.
- W4287748764 hasPrimaryLocation W42877487641 @default.
- W4287748764 hasRelatedWork W1562959674 @default.
- W4287748764 hasRelatedWork W2099126390 @default.
- W4287748764 hasRelatedWork W2323035519 @default.
- W4287748764 hasRelatedWork W3035515490 @default.
- W4287748764 hasRelatedWork W3038896498 @default.
- W4287748764 hasRelatedWork W3074294383 @default.
- W4287748764 hasRelatedWork W3173482257 @default.
- W4287748764 hasRelatedWork W3209094908 @default.
- W4287748764 hasRelatedWork W4210912933 @default.
- W4287748764 hasRelatedWork W4287748764 @default.
- W4287748764 isParatext "false" @default.
- W4287748764 isRetracted "false" @default.
- W4287748764 workType "article" @default.