Matches in SemOpenAlex for { <https://semopenalex.org/work/W4385571602> ?p ?o ?g. }
Showing items 1 to 65 of
65
with 100 items per page.
- W4385571602 abstract "Natural language video localization(NLVL) task involves the semantic matching of a text query with a moment from an untrimmed video. Previous methods primarily focus on improving performance with the assumption of independently identical data distribution while ignoring the out-of-distribution data. Therefore, these approaches often fail when handling the videos and queries in novel scenes, which is inevitable in real-world scenarios. In this paper, we, for the first time, formulate the scene-robust NLVL problem and propose a novel generalizable NLVL framework utilizing data in multiple available scenes to learn a robust model. Specifically, our model learns a group of generalizable domain-invariant representations by alignment and decomposition. First, we propose a comprehensive intra- and inter-sample distance metric for complex multi-modal feature space, and an asymmetric multi-modal alignment loss for different information densities of text and vision. Further, to alleviate the conflict between domain-invariant features for generalization and domain-specific information for reasoning, we introduce domain-specific and domain-agnostic predictors to decompose and refine the learned features by dynamically adjusting the weights of samples. Based on the original video tags, we conduct extensive experiments on three NLVL datasets with different-grained scene shifts to show the effectiveness of our proposed methods." @default.
- W4385571602 created "2023-08-05" @default.
- W4385571602 creator A5010433680 @default.
- W4385571602 creator A5012728201 @default.
- W4385571602 creator A5028127027 @default.
- W4385571602 creator A5031105985 @default.
- W4385571602 creator A5065361552 @default.
- W4385571602 date "2023-01-01" @default.
- W4385571602 modified "2023-10-18" @default.
- W4385571602 title "Scene-robust Natural Language Video Localization via Learning Domain-invariant Representations" @default.
- W4385571602 doi "https://doi.org/10.18653/v1/2023.findings-acl.11" @default.
- W4385571602 hasPublicationYear "2023" @default.
- W4385571602 type Work @default.
- W4385571602 citedByCount "0" @default.
- W4385571602 crossrefType "proceedings-article" @default.
- W4385571602 hasAuthorship W4385571602A5010433680 @default.
- W4385571602 hasAuthorship W4385571602A5012728201 @default.
- W4385571602 hasAuthorship W4385571602A5028127027 @default.
- W4385571602 hasAuthorship W4385571602A5031105985 @default.
- W4385571602 hasAuthorship W4385571602A5065361552 @default.
- W4385571602 hasBestOaLocation W43855716021 @default.
- W4385571602 hasConcept C105795698 @default.
- W4385571602 hasConcept C119857082 @default.
- W4385571602 hasConcept C134306372 @default.
- W4385571602 hasConcept C153180895 @default.
- W4385571602 hasConcept C154945302 @default.
- W4385571602 hasConcept C165064840 @default.
- W4385571602 hasConcept C177148314 @default.
- W4385571602 hasConcept C190470478 @default.
- W4385571602 hasConcept C195324797 @default.
- W4385571602 hasConcept C31972630 @default.
- W4385571602 hasConcept C33923547 @default.
- W4385571602 hasConcept C36503486 @default.
- W4385571602 hasConcept C37914503 @default.
- W4385571602 hasConcept C41008148 @default.
- W4385571602 hasConceptScore W4385571602C105795698 @default.
- W4385571602 hasConceptScore W4385571602C119857082 @default.
- W4385571602 hasConceptScore W4385571602C134306372 @default.
- W4385571602 hasConceptScore W4385571602C153180895 @default.
- W4385571602 hasConceptScore W4385571602C154945302 @default.
- W4385571602 hasConceptScore W4385571602C165064840 @default.
- W4385571602 hasConceptScore W4385571602C177148314 @default.
- W4385571602 hasConceptScore W4385571602C190470478 @default.
- W4385571602 hasConceptScore W4385571602C195324797 @default.
- W4385571602 hasConceptScore W4385571602C31972630 @default.
- W4385571602 hasConceptScore W4385571602C33923547 @default.
- W4385571602 hasConceptScore W4385571602C36503486 @default.
- W4385571602 hasConceptScore W4385571602C37914503 @default.
- W4385571602 hasConceptScore W4385571602C41008148 @default.
- W4385571602 hasLocation W43855716021 @default.
- W4385571602 hasOpenAccess W4385571602 @default.
- W4385571602 hasPrimaryLocation W43855716021 @default.
- W4385571602 hasRelatedWork W1891287906 @default.
- W4385571602 hasRelatedWork W1995188412 @default.
- W4385571602 hasRelatedWork W2025991752 @default.
- W4385571602 hasRelatedWork W2391245565 @default.
- W4385571602 hasRelatedWork W2533072256 @default.
- W4385571602 hasRelatedWork W2544283655 @default.
- W4385571602 hasRelatedWork W2786306966 @default.
- W4385571602 hasRelatedWork W2961085424 @default.
- W4385571602 hasRelatedWork W3111677651 @default.
- W4385571602 hasRelatedWork W4306674287 @default.
- W4385571602 isParatext "false" @default.
- W4385571602 isRetracted "false" @default.
- W4385571602 workType "article" @default.