Re-evaluating Group Robustness via Adaptive Class-Specific Scaling

I tiakina i:
Ngā taipitopito rārangi puna kōrero
I whakaputaina i:arXiv.org (Dec 19, 2024), p. n/a
Kaituhi matua: Seo, Seonguk
Ētahi atu kaituhi: Han, Bohyung
I whakaputaina:
Cornell University Library, arXiv.org
Ngā marau:
Urunga tuihono:Citation/Abstract
Full text outside of ProQuest
Ngā Tūtohu: Tāpirihia he Tūtohu
Kāore He Tūtohu, Me noho koe te mea tuatahi ki te tūtohu i tēnei pūkete!

MARC

LEADER 00000nab a2200000uu 4500
001 3148683248
003 UK-CbPIL
022 |a 2331-8422 
035 |a 3148683248 
045 0 |b d20241219 
100 1 |a Seo, Seonguk 
245 1 |a Re-evaluating Group Robustness via Adaptive Class-Specific Scaling 
260 |b Cornell University Library, arXiv.org  |c Dec 19, 2024 
513 |a Working Paper 
520 3 |a Group distributionally robust optimization, which aims to improve robust accuracies -- worst-group and unbiased accuracies -- is a prominent algorithm used to mitigate spurious correlations and address dataset bias. Although existing approaches have reported improvements in robust accuracies, these gains often come at the cost of average accuracy due to inherent trade-offs. To control this trade-off flexibly and efficiently, we propose a simple class-specific scaling strategy, directly applicable to existing debiasing algorithms with no additional training. We further develop an instance-wise adaptive scaling technique to alleviate this trade-off, even leading to improvements in both robust and average accuracies. Our approach reveals that a na\"ive ERM baseline matches or even outperforms the recent debiasing methods by simply adopting the class-specific scaling technique. Additionally, we introduce a novel unified metric that quantifies the trade-off between the two accuracies as a scalar value, allowing for a comprehensive evaluation of existing algorithms. By tackling the inherent trade-off and offering a performance landscape, our approach provides valuable insights into robust techniques beyond just robust accuracy. We validate the effectiveness of our framework through experiments across datasets in computer vision and natural language processing domains. 
653 |a Scaling 
653 |a Datasets 
653 |a Robust control 
653 |a Computer vision 
653 |a Performance evaluation 
653 |a Algorithms 
653 |a Natural language processing 
653 |a Tradeoffs 
653 |a Adaptive algorithms 
700 1 |a Han, Bohyung 
773 0 |t arXiv.org  |g (Dec 19, 2024), p. n/a 
786 0 |d ProQuest  |t Engineering Database 
856 4 1 |3 Citation/Abstract  |u https://www.proquest.com/docview/3148683248/abstract/embedded/6A8EOT78XXH2IG52?source=fedsrch 
856 4 0 |3 Full text outside of ProQuest  |u http://arxiv.org/abs/2412.15311