Matches in SemOpenAlex for { <https://semopenalex.org/work/W4320473147> ?p ?o ?g. }
- W4320473147 endingPage "319" @default.
- W4320473147 startingPage "307" @default.
- W4320473147 abstract "Deep neural networks easily suffer from weak generalization caused by overfitting on biased data. One popular remedy to alleviate this issue is sample reweighting methods that adaptively adjust the importance of biased samples. Separate from the effort to reduce bias, recent works show that the generalization power can be improved by auxiliary tasks. Inspired by the two lines of works, we extend the sample reweighting methods to auxiliary tasks. In this paper, we propose a novel auxiliary learning framework that improves the primary task by adaptively adjusting the weights of samples from multiple tasks rather than samples from a single task using a weighting function. The weighting function is optimized by meta-learning along the gradient of the loss for meta-data, which is a small unbiased validation data. We also present a task-activation score that indicates the correlation between the learning tendency of the training samples and meta-data samples. This score is utilized as a regularizer for meta-learning objective. Our framework can obtain powerful representations for the primary task on biased data by automatically identifying effective combinations of tasks. Our experiments demonstrate that our proposed method consistently outperforms all baselines and state-of-the-art methods on both corrupted labels and class imbalance settings." @default.
- W4320473147 created "2023-02-14" @default.
- W4320473147 creator A5012207088 @default.
- W4320473147 creator A5041112881 @default.
- W4320473147 creator A5051115125 @default.
- W4320473147 creator A5063090617 @default.
- W4320473147 creator A5084814930 @default.
- W4320473147 date "2023-05-01" @default.
- W4320473147 modified "2023-09-26" @default.
- W4320473147 title "Robust auxiliary learning with weighting function for biased data" @default.
- W4320473147 cites W1988790447 @default.
- W4320473147 cites W2052250245 @default.
- W4320473147 cites W2118978333 @default.
- W4320473147 cites W2122646361 @default.
- W4320473147 cites W2165698076 @default.
- W4320473147 cites W2194775991 @default.
- W4320473147 cites W2256388387 @default.
- W4320473147 cites W2557912764 @default.
- W4320473147 cites W2565516711 @default.
- W4320473147 cites W2944842185 @default.
- W4320473147 cites W2963351448 @default.
- W4320473147 cites W2963516811 @default.
- W4320473147 cites W2963691377 @default.
- W4320473147 cites W2963877604 @default.
- W4320473147 cites W2979302305 @default.
- W4320473147 cites W2992308087 @default.
- W4320473147 cites W3046688540 @default.
- W4320473147 cites W3110687497 @default.
- W4320473147 cites W3202087803 @default.
- W4320473147 cites W3217201372 @default.
- W4320473147 cites W4221161368 @default.
- W4320473147 cites W4284680369 @default.
- W4320473147 doi "https://doi.org/10.1016/j.ins.2023.01.099" @default.
- W4320473147 hasPublicationYear "2023" @default.
- W4320473147 type Work @default.
- W4320473147 citedByCount "0" @default.
- W4320473147 crossrefType "journal-article" @default.
- W4320473147 hasAuthorship W4320473147A5012207088 @default.
- W4320473147 hasAuthorship W4320473147A5041112881 @default.
- W4320473147 hasAuthorship W4320473147A5051115125 @default.
- W4320473147 hasAuthorship W4320473147A5063090617 @default.
- W4320473147 hasAuthorship W4320473147A5084814930 @default.
- W4320473147 hasBestOaLocation W43204731471 @default.
- W4320473147 hasConcept C119857082 @default.
- W4320473147 hasConcept C126838900 @default.
- W4320473147 hasConcept C134306372 @default.
- W4320473147 hasConcept C14036430 @default.
- W4320473147 hasConcept C153180895 @default.
- W4320473147 hasConcept C154945302 @default.
- W4320473147 hasConcept C162324750 @default.
- W4320473147 hasConcept C177148314 @default.
- W4320473147 hasConcept C183115368 @default.
- W4320473147 hasConcept C185592680 @default.
- W4320473147 hasConcept C187736073 @default.
- W4320473147 hasConcept C198531522 @default.
- W4320473147 hasConcept C22019652 @default.
- W4320473147 hasConcept C2780451532 @default.
- W4320473147 hasConcept C2781002164 @default.
- W4320473147 hasConcept C28006648 @default.
- W4320473147 hasConcept C33923547 @default.
- W4320473147 hasConcept C41008148 @default.
- W4320473147 hasConcept C43617362 @default.
- W4320473147 hasConcept C50644808 @default.
- W4320473147 hasConcept C71924100 @default.
- W4320473147 hasConcept C78458016 @default.
- W4320473147 hasConcept C86803240 @default.
- W4320473147 hasConceptScore W4320473147C119857082 @default.
- W4320473147 hasConceptScore W4320473147C126838900 @default.
- W4320473147 hasConceptScore W4320473147C134306372 @default.
- W4320473147 hasConceptScore W4320473147C14036430 @default.
- W4320473147 hasConceptScore W4320473147C153180895 @default.
- W4320473147 hasConceptScore W4320473147C154945302 @default.
- W4320473147 hasConceptScore W4320473147C162324750 @default.
- W4320473147 hasConceptScore W4320473147C177148314 @default.
- W4320473147 hasConceptScore W4320473147C183115368 @default.
- W4320473147 hasConceptScore W4320473147C185592680 @default.
- W4320473147 hasConceptScore W4320473147C187736073 @default.
- W4320473147 hasConceptScore W4320473147C198531522 @default.
- W4320473147 hasConceptScore W4320473147C22019652 @default.
- W4320473147 hasConceptScore W4320473147C2780451532 @default.
- W4320473147 hasConceptScore W4320473147C2781002164 @default.
- W4320473147 hasConceptScore W4320473147C28006648 @default.
- W4320473147 hasConceptScore W4320473147C33923547 @default.
- W4320473147 hasConceptScore W4320473147C41008148 @default.
- W4320473147 hasConceptScore W4320473147C43617362 @default.
- W4320473147 hasConceptScore W4320473147C50644808 @default.
- W4320473147 hasConceptScore W4320473147C71924100 @default.
- W4320473147 hasConceptScore W4320473147C78458016 @default.
- W4320473147 hasConceptScore W4320473147C86803240 @default.
- W4320473147 hasLocation W43204731471 @default.
- W4320473147 hasOpenAccess W4320473147 @default.
- W4320473147 hasPrimaryLocation W43204731471 @default.
- W4320473147 hasRelatedWork W3040863728 @default.
- W4320473147 hasRelatedWork W3092824172 @default.
- W4320473147 hasRelatedWork W3093147807 @default.
- W4320473147 hasRelatedWork W3105010138 @default.
- W4320473147 hasRelatedWork W3105036711 @default.
- W4320473147 hasRelatedWork W3112642831 @default.