Last layer re-training is sufficient for robustness to spurious correlations

P Kirichenko, P Izmailov, AG Wilson - arXiv preprint arXiv:2204.02937, 2022 - arxiv.org
Neural network classifiers can largely rely on simple spurious features, such as
backgrounds, to make predictions. However, even in these cases, we show that they still …
被引用次数：212 相关文章

Last Layer Re-Training is Sufficient for Robustness to Spurious Correlations

P Kirichenko, P Izmailov, AG Wilson - The Eleventh International … - openreview.net
Neural network classifiers can largely rely on simple spurious features, such as image
backgrounds, to make predictions. However, even in these cases, we show that they still …

Last Layer Re-Training is Sufficient for Robustness to Spurious Correlations

P Kirichenko, P Izmailov, AG Wilson - International Conference on …, 2022 - par.nsf.gov
Neural network classifiers can largely rely on simple spurious features, such as
backgrounds, to make predictions. However, even in these cases, we show that they still …

Last Layer Re-Training is Sufficient for Robustness to Spurious Correlations

P Kirichenko, P Izmailov, AG Wilson - arXiv e-prints, 2022 - ui.adsabs.harvard.edu
Neural network classifiers can largely rely on simple spurious features, such as
backgrounds, to make predictions. However, even in these cases, we show that they still …

Last Layer Re-Training is Sufficient for Robustness to Spurious Correlations

P Kirichenko, P Izmailov, AG Wilson - ICML 2022: Workshop on Spurious … - openreview.net
Neural network classifiers can largely rely on simple spurious features, such as
backgrounds, to make predictions. However, even in these cases, we show that they still …

Last Layer Re-Training is Sufficient for Robustness to Spurious Correlations

P Kirichenko, P Izmailov, A Gordon Wilson - ICLR 2023, 2023 - par.nsf.gov
Neural network classifiers can largely rely on simple spurious features, such as
backgrounds, to make predictions. However, even in these cases, we show that they still …