Matches in SemOpenAlex for { <https://semopenalex.org/work/W3204672260> ?p ?o ?g. }
- W3204672260 abstract "While neural symbolic methods demonstrate impressive performance in visual question answering on synthetic images, their performance suffers on real images. We identify that the long-tail distribution of visual concepts and unequal importance of reasoning steps in real data are the two key obstacles that limit the models' real-world potentials. To address these challenges, we propose a new paradigm, Calibrating Concepts and Operations (CCO), which enables neural symbolic models to capture underlying data characteristics and to reason with hierarchical importance. Specifically, we introduce an executor with learnable concept embedding magnitudes for handling distribution imbalance, and an operation calibrator for highlighting important operations and suppressing redundant ones. Our experiments show CCO substantially boosts the performance of neural symbolic methods on real images. By evaluating models on the real world dataset GQA, CCO helps the neural symbolic method NSCL outperforms its vanilla counterpart by 9.1% (from 47.0% to 56.1%); this result also largely reduces the performance gap between symbolic and non-symbolic methods. Additionally, we create a perturbed test set for better understanding and analyzing model performance on real images. Code is available at https://github.com/Lizw14/CaliCO.git ." @default.
- W3204672260 created "2021-10-11" @default.
- W3204672260 creator A5004491730 @default.
- W3204672260 creator A5009286723 @default.
- W3204672260 creator A5017658063 @default.
- W3204672260 creator A5022344556 @default.
- W3204672260 creator A5038276570 @default.
- W3204672260 creator A5075825791 @default.
- W3204672260 creator A5086706224 @default.
- W3204672260 date "2021-10-01" @default.
- W3204672260 modified "2023-09-27" @default.
- W3204672260 title "Calibrating Concepts and Operations: Towards Symbolic Reasoning on Real Images" @default.
- W3204672260 cites W1514535095 @default.
- W3204672260 cites W1933349210 @default.
- W3204672260 cites W2250539671 @default.
- W3204672260 cites W2277195237 @default.
- W3204672260 cites W2410323755 @default.
- W3204672260 cites W2412400526 @default.
- W3204672260 cites W2463267937 @default.
- W3204672260 cites W2560730294 @default.
- W3204672260 cites W2561715562 @default.
- W3204672260 cites W2575842049 @default.
- W3204672260 cites W2745461083 @default.
- W3204672260 cites W2786209943 @default.
- W3204672260 cites W2804243936 @default.
- W3204672260 cites W2883104598 @default.
- W3204672260 cites W2890952061 @default.
- W3204672260 cites W2891021031 @default.
- W3204672260 cites W2904910963 @default.
- W3204672260 cites W2951391755 @default.
- W3204672260 cites W2953106684 @default.
- W3204672260 cites W2953388933 @default.
- W3204672260 cites W2955124656 @default.
- W3204672260 cites W2958447893 @default.
- W3204672260 cites W2962716332 @default.
- W3204672260 cites W2963224792 @default.
- W3204672260 cites W2963518342 @default.
- W3204672260 cites W2963630207 @default.
- W3204672260 cites W2963644680 @default.
- W3204672260 cites W2963690694 @default.
- W3204672260 cites W2963954913 @default.
- W3204672260 cites W2964118342 @default.
- W3204672260 cites W2964345792 @default.
- W3204672260 cites W2970231061 @default.
- W3204672260 cites W2970796523 @default.
- W3204672260 cites W2971200848 @default.
- W3204672260 cites W2982699810 @default.
- W3204672260 cites W2986755220 @default.
- W3204672260 cites W2998088552 @default.
- W3204672260 cites W3034653874 @default.
- W3204672260 cites W3034682719 @default.
- W3204672260 cites W3035393139 @default.
- W3204672260 cites W3035517717 @default.
- W3204672260 cites W3110444491 @default.
- W3204672260 cites W3118500473 @default.
- W3204672260 cites W3176187895 @default.
- W3204672260 cites W3177225079 @default.
- W3204672260 doi "https://doi.org/10.48550/arxiv.2110.00519" @default.
- W3204672260 hasPublicationYear "2021" @default.
- W3204672260 type Work @default.
- W3204672260 sameAs 3204672260 @default.
- W3204672260 citedByCount "0" @default.
- W3204672260 crossrefType "posted-content" @default.
- W3204672260 hasAuthorship W3204672260A5004491730 @default.
- W3204672260 hasAuthorship W3204672260A5009286723 @default.
- W3204672260 hasAuthorship W3204672260A5017658063 @default.
- W3204672260 hasAuthorship W3204672260A5022344556 @default.
- W3204672260 hasAuthorship W3204672260A5038276570 @default.
- W3204672260 hasAuthorship W3204672260A5075825791 @default.
- W3204672260 hasAuthorship W3204672260A5086706224 @default.
- W3204672260 hasBestOaLocation W32046722601 @default.
- W3204672260 hasConcept C11171543 @default.
- W3204672260 hasConcept C119857082 @default.
- W3204672260 hasConcept C154945302 @default.
- W3204672260 hasConcept C15744967 @default.
- W3204672260 hasConcept C177264268 @default.
- W3204672260 hasConcept C17744445 @default.
- W3204672260 hasConcept C180591056 @default.
- W3204672260 hasConcept C199360897 @default.
- W3204672260 hasConcept C199539241 @default.
- W3204672260 hasConcept C26517878 @default.
- W3204672260 hasConcept C2776095079 @default.
- W3204672260 hasConcept C2776760102 @default.
- W3204672260 hasConcept C38652104 @default.
- W3204672260 hasConcept C41008148 @default.
- W3204672260 hasConcept C41608201 @default.
- W3204672260 hasConcept C50644808 @default.
- W3204672260 hasConceptScore W3204672260C11171543 @default.
- W3204672260 hasConceptScore W3204672260C119857082 @default.
- W3204672260 hasConceptScore W3204672260C154945302 @default.
- W3204672260 hasConceptScore W3204672260C15744967 @default.
- W3204672260 hasConceptScore W3204672260C177264268 @default.
- W3204672260 hasConceptScore W3204672260C17744445 @default.
- W3204672260 hasConceptScore W3204672260C180591056 @default.
- W3204672260 hasConceptScore W3204672260C199360897 @default.
- W3204672260 hasConceptScore W3204672260C199539241 @default.
- W3204672260 hasConceptScore W3204672260C26517878 @default.
- W3204672260 hasConceptScore W3204672260C2776095079 @default.
- W3204672260 hasConceptScore W3204672260C2776760102 @default.
- W3204672260 hasConceptScore W3204672260C38652104 @default.