Matches in SemOpenAlex for { <https://semopenalex.org/work/W3198849614> ?p ?o ?g. }
- W3198849614 abstract "Machine learning models achieve state-of-the-art performance on many supervised learning tasks. However, prior evidence suggests that these models may learn to rely on shortcut biases or spurious correlations (intuitively, correlations that do not hold in the test as they hold in train) for good predictive performance. Such models cannot be trusted in deployment environments to provide accurate predictions. While viewing the problem from a causal lens is known to be useful, the seamless integration of causation techniques into machine learning pipelines remains cumbersome and expensive. In this work, we study and extend a causal pre-training debiasing technique called causal bootstrapping (CB) under five practical confounded-data generation-acquisition scenarios (with known and unknown confounding). Under these settings, we systematically investigate the effect of confounding bias on deep learning model performance, demonstrating their propensity to rely on shortcut biases when these biases are not properly accounted for. We demonstrate that such a causal pre-training technique can significantly outperform existing base practices to mitigate confounding bias on real-world domain generalization benchmarking tasks. This systematic investigation underlines the importance of accounting for the underlying data-generating mechanisms and fortifying data-preprocessing pipelines with a causal framework to develop methods robust to confounding biases." @default.
- W3198849614 created "2021-09-13" @default.
- W3198849614 creator A5014929343 @default.
- W3198849614 creator A5019357628 @default.
- W3198849614 creator A5035149567 @default.
- W3198849614 creator A5070063054 @default.
- W3198849614 date "2021-08-27" @default.
- W3198849614 modified "2023-09-25" @default.
- W3198849614 title "Pulling Up by the Causal Bootstraps: Causal Data Augmentation for Pre-training Debiasing" @default.
- W3198849614 cites W1486066286 @default.
- W3198849614 cites W1537314526 @default.
- W3198849614 cites W1834627138 @default.
- W3198849614 cites W1905064697 @default.
- W3198849614 cites W1978380814 @default.
- W3198849614 cites W2030498706 @default.
- W3198849614 cites W2049910836 @default.
- W3198849614 cites W2131953535 @default.
- W3198849614 cites W2134067266 @default.
- W3198849614 cites W2143891888 @default.
- W3198849614 cites W2162670686 @default.
- W3198849614 cites W2299453729 @default.
- W3198849614 cites W2342840547 @default.
- W3198849614 cites W2557738935 @default.
- W3198849614 cites W2580744997 @default.
- W3198849614 cites W2581082771 @default.
- W3198849614 cites W2605072939 @default.
- W3198849614 cites W2616901848 @default.
- W3198849614 cites W2740962769 @default.
- W3198849614 cites W2774616426 @default.
- W3198849614 cites W2788481061 @default.
- W3198849614 cites W2788557041 @default.
- W3198849614 cites W2790376986 @default.
- W3198849614 cites W2811374795 @default.
- W3198849614 cites W2825122211 @default.
- W3198849614 cites W2882983190 @default.
- W3198849614 cites W2883386984 @default.
- W3198849614 cites W2887175137 @default.
- W3198849614 cites W2889232360 @default.
- W3198849614 cites W2912457762 @default.
- W3198849614 cites W2937229771 @default.
- W3198849614 cites W2949505205 @default.
- W3198849614 cites W2952428244 @default.
- W3198849614 cites W2952610664 @default.
- W3198849614 cites W2953494151 @default.
- W3198849614 cites W2963350032 @default.
- W3198849614 cites W2963453204 @default.
- W3198849614 cites W2963466845 @default.
- W3198849614 cites W2963608118 @default.
- W3198849614 cites W2965628639 @default.
- W3198849614 cites W2970990331 @default.
- W3198849614 cites W2971274354 @default.
- W3198849614 cites W2981869278 @default.
- W3198849614 cites W2990751682 @default.
- W3198849614 cites W2995225687 @default.
- W3198849614 cites W3005040148 @default.
- W3198849614 cites W3007247345 @default.
- W3198849614 cites W3011762034 @default.
- W3198849614 cites W3011967257 @default.
- W3198849614 cites W3027716283 @default.
- W3198849614 cites W3035037113 @default.
- W3198849614 cites W3041475347 @default.
- W3198849614 cites W3044799724 @default.
- W3198849614 cites W3049256904 @default.
- W3198849614 cites W3101156210 @default.
- W3198849614 cites W3112486745 @default.
- W3198849614 cites W3159623667 @default.
- W3198849614 cites W3172100757 @default.
- W3198849614 cites W102511605 @default.
- W3198849614 hasPublicationYear "2021" @default.
- W3198849614 type Work @default.
- W3198849614 sameAs 3198849614 @default.
- W3198849614 citedByCount "0" @default.
- W3198849614 crossrefType "posted-content" @default.
- W3198849614 hasAuthorship W3198849614A5014929343 @default.
- W3198849614 hasAuthorship W3198849614A5019357628 @default.
- W3198849614 hasAuthorship W3198849614A5035149567 @default.
- W3198849614 hasAuthorship W3198849614A5070063054 @default.
- W3198849614 hasConcept C105795698 @default.
- W3198849614 hasConcept C119857082 @default.
- W3198849614 hasConcept C124101348 @default.
- W3198849614 hasConcept C149782125 @default.
- W3198849614 hasConcept C154945302 @default.
- W3198849614 hasConcept C15744967 @default.
- W3198849614 hasConcept C158600405 @default.
- W3198849614 hasConcept C166151441 @default.
- W3198849614 hasConcept C17744445 @default.
- W3198849614 hasConcept C188147891 @default.
- W3198849614 hasConcept C199539241 @default.
- W3198849614 hasConcept C207609745 @default.
- W3198849614 hasConcept C2779458634 @default.
- W3198849614 hasConcept C33923547 @default.
- W3198849614 hasConcept C41008148 @default.
- W3198849614 hasConcept C77350462 @default.
- W3198849614 hasConcept C97256817 @default.
- W3198849614 hasConceptScore W3198849614C105795698 @default.
- W3198849614 hasConceptScore W3198849614C119857082 @default.
- W3198849614 hasConceptScore W3198849614C124101348 @default.
- W3198849614 hasConceptScore W3198849614C149782125 @default.
- W3198849614 hasConceptScore W3198849614C154945302 @default.
- W3198849614 hasConceptScore W3198849614C15744967 @default.