FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4309619902> ?p ?o ?g. }

• W4309619902 endingPage "28" @default.
• W4309619902 startingPage "1" @default.
• W4309619902 abstract "While Deep Neural Networks (DNNs) are deriving the major innovations through their powerful automation, we are also witnessing the peril behind automation as a form of bias, such as automated racism, gender bias, and adversarial bias. As the societal impact of DNNs grows, finding an effective way to steer DNNs to align their behavior with the human mental model has become indispensable in realizing fair and accountable models. While establishing the way to adjust DNNs to think like humans'' is in pressing need, there have been few approaches aiming to capture how humans would think'' when DNNs introduce biased reasoning in seeing a new instance. We propose Interactive Attention Alignment (IAA), a framework that uses the methods for visualizing model attention, such as saliency maps, as an interactive medium that humans can leverage to unveil the cases of DNN's biased reasoning and directly adjust the attention. To realize more effective human-steerable DNNs than state-of-the-art, IAA introduces two novel devices. First, IAA uses Reasonability Matrix to systematically identify and adjust the cases of biased attention. Second, IAA applies GRADIA, a computational pipeline designed for effectively applying the adjusted attention to jointly maximize attention quality and prediction accuracy. We evaluated Reasonability Matrix in Study 1 and GRADIA in Study 2 in the gender classification problem. In Study 1, we found applying Reasonability Matrix in bias detection can significantly improve the perceived quality of model attention from human eyes than not applying Reasonability Matrix. In Study 2, we found using GRADIA significantly improves (1) the human-assessed perceived quality of model attention and (2) model performance in scenarios where the training samples are limited. Based on our observation in the two studies, we present implications for future design in the problem space of social computing and interactive data annotation toward achieving a human-centered steerable AI." @default.
• W4309619902 created "2022-11-28" @default.
• W4309619902 creator A5042962342 @default.
• W4309619902 creator A5048756500 @default.
• W4309619902 creator A5057999830 @default.
• W4309619902 creator A5076632029 @default.
• W4309619902 date "2022-11-07" @default.
• W4309619902 modified "2023-09-27" @default.
• W4309619902 title "Aligning Eyes between Humans and Deep Neural Network through Interactive Attention Alignment" @default.
• W4309619902 cites W1861492603 @default.
• W4309619902 cites W2003238113 @default.
• W4309619902 cites W2012457675 @default.
• W4309619902 cites W2026019770 @default.
• W4309619902 cites W2080417696 @default.
• W4309619902 cites W2097246321 @default.
• W4309619902 cites W2116666691 @default.
• W4309619902 cites W2118978333 @default.
• W4309619902 cites W2127058057 @default.
• W4309619902 cites W2133990480 @default.
• W4309619902 cites W2137406659 @default.
• W4309619902 cites W2148143831 @default.
• W4309619902 cites W2150997454 @default.
• W4309619902 cites W2157928966 @default.
• W4309619902 cites W2175342987 @default.
• W4309619902 cites W2186022498 @default.
• W4309619902 cites W2194775991 @default.
• W4309619902 cites W2295107390 @default.
• W4309619902 cites W2394669110 @default.
• W4309619902 cites W2440722286 @default.
• W4309619902 cites W2583689529 @default.
• W4309619902 cites W2607223307 @default.
• W4309619902 cites W2740693122 @default.
• W4309619902 cites W2759653627 @default.
• W4309619902 cites W2781228439 @default.
• W4309619902 cites W2804927761 @default.
• W4309619902 cites W2883424428 @default.
• W4309619902 cites W2889730816 @default.
• W4309619902 cites W2904239671 @default.
• W4309619902 cites W2937229771 @default.
• W4309619902 cites W2942161347 @default.
• W4309619902 cites W2962772482 @default.
• W4309619902 cites W2962858109 @default.
• W4309619902 cites W2962884579 @default.
• W4309619902 cites W2963082289 @default.
• W4309619902 cites W2963350032 @default.
• W4309619902 cites W2963588812 @default.
• W4309619902 cites W2963749936 @default.
• W4309619902 cites W2964200170 @default.
• W4309619902 cites W2964286876 @default.
• W4309619902 cites W2990751682 @default.
• W4309619902 cites W2998014937 @default.
• W4309619902 cites W3006437051 @default.
• W4309619902 cites W3012736183 @default.
• W4309619902 cites W3019489177 @default.
• W4309619902 cites W3031923829 @default.
• W4309619902 cites W3106426003 @default.
• W4309619902 cites W3119150429 @default.
• W4309619902 cites W3128601380 @default.
• W4309619902 cites W3156354433 @default.
• W4309619902 cites W3167947413 @default.
• W4309619902 cites W3206015227 @default.
• W4309619902 cites W4206952613 @default.
• W4309619902 cites W4283703617 @default.
• W4309619902 cites W4283824940 @default.
• W4309619902 doi "https://doi.org/10.1145/3555590" @default.
• W4309619902 hasPublicationYear "2022" @default.
• W4309619902 type Work @default.
• W4309619902 citedByCount "4" @default.
• W4309619902 countsByYear W43096199022022 @default.
• W4309619902 countsByYear W43096199022023 @default.
• W4309619902 crossrefType "journal-article" @default.
• W4309619902 hasAuthorship W4309619902A5042962342 @default.
• W4309619902 hasAuthorship W4309619902A5048756500 @default.
• W4309619902 hasAuthorship W4309619902A5057999830 @default.
• W4309619902 hasAuthorship W4309619902A5076632029 @default.
• W4309619902 hasBestOaLocation W43096199022 @default.
• W4309619902 hasConcept C106487976 @default.
• W4309619902 hasConcept C115901376 @default.
• W4309619902 hasConcept C119857082 @default.
• W4309619902 hasConcept C127413603 @default.
• W4309619902 hasConcept C153083717 @default.
• W4309619902 hasConcept C154945302 @default.
• W4309619902 hasConcept C159985019 @default.
• W4309619902 hasConcept C192562407 @default.
• W4309619902 hasConcept C2984842247 @default.
• W4309619902 hasConcept C41008148 @default.
• W4309619902 hasConcept C50644808 @default.
• W4309619902 hasConcept C78519656 @default.
• W4309619902 hasConceptScore W4309619902C106487976 @default.
• W4309619902 hasConceptScore W4309619902C115901376 @default.
• W4309619902 hasConceptScore W4309619902C119857082 @default.
• W4309619902 hasConceptScore W4309619902C127413603 @default.
• W4309619902 hasConceptScore W4309619902C153083717 @default.
• W4309619902 hasConceptScore W4309619902C154945302 @default.
• W4309619902 hasConceptScore W4309619902C159985019 @default.
• W4309619902 hasConceptScore W4309619902C192562407 @default.
• W4309619902 hasConceptScore W4309619902C2984842247 @default.
• W4309619902 hasConceptScore W4309619902C41008148 @default.

• next