paper-with-me

홈 › Papers

RLSbench: Domain Adaptation Under Relaxed Label Shift

2023-02-06 · Saurabh Garg, Nick Erickson, James Sharpnack, Alex Smola, Sivaraman Balakrishnan, Zachary C. Lipton

Despite the emergence of principled methods for domain adaptation under label shift, their sensitivity to shifts in class conditional distributions is precariously under explored. Meanwhile, popular deep domain adaptation heuristics tend to falter when faced with label proportions shifts. While several papers modify these heuristics in attempts to handle label proportions shifts, inconsistencies in evaluation standards, datasets, and baselines make it difficult to gauge the current best practices. In this paper, we introduce RLSbench, a large-scale benchmark for relaxed label shift, consisting of $>$500 distribution shift pairs spanning vision, tabular, and language modalities, with varying label proportions. Unlike existing benchmarks, which primarily focus on shifts in class-conditional $p(x|y)$, our benchmark also focuses on label marginal shifts. First, we assess 13 popular domain adaptation methods, demonstrating more widespread failures under label proportion shifts than were previously known. Next, we develop an effective two-step meta-algorithm that is compatible with most domain adaptation heuristics: (i) pseudo-balance the data at each epoch; and (ii) adjust the final classifier with target label distribution estimate. The meta-algorithm improves existing domain adaptation heuristics under large label proportion shifts, often by 2--10\% accuracy points, while conferring minimal effect ($<$0.5\%) when label proportions do not shift. We hope that these findings and the availability of RLSbench will encourage researchers to rigorously evaluate proposed methods in relaxed label shift settings. Code is publicly available at https://github.com/acmi-lab/RLSbench.

📄 PDF Abstract BibTeX arXiv:2302.03020

Code (1)

acmi-lab/rlsbench 공식 구현 pytorch

Tasks

Domain Adaptation

Similar Papers 제목 키워드 기반

Domain Adaptation with Asymmetrically-Relaxed Distribution Alignment

2019-03-05 · ICLR Workshop LLD 2019 · Yifan Wu, Ezra Winston, Divyansh Kaushik, Zachary Lipton

Domain adaptation addresses the common problem when the target distribution generating our test data drifts from the source (training) distribution. While absent assumptions, domain adaptation is impossible, strict condi…

Domain Adaptation

Connecting sufficient conditions for domain adaptation: source-guided uncertainty, relaxed divergences and discrepancy localization

2022-03-09 · Sofien Dhouib, Setareh Maghsudi

Recent advances in domain adaptation establish that requiring a low risk on the source domain and equal feature marginals degrade the adaptation's performance. At the same time, empirical evidence shows that incorporatin…

Domain Adaptation

Relaxed Conditional Image Transfer for Semi-supervised Domain Adaptation

2021-01-05 · Qijun Luo, Zhili Liu, Lanqing Hong, Chongxuan Li 외

Semi-supervised domain adaptation (SSDA), which aims to learn models in a partially labeled target domain with the assistance of the fully labeled source domain, attracts increasing attention in recent years. To explicit…

Domain AdaptationSemi-supervised Domain Adaptation

Memory Consistent Unsupervised Off-the-Shelf Model Adaptation for Source-Relaxed Medical Image Segmentation

2022-09-16 · Xiaofeng Liu, Fangxu Xing, Georges El Fakhri, Jonghye Woo

Unsupervised domain adaptation (UDA) has been a vital protocol for migrating information learned from a labeled source domain to facilitate the implementation in an unlabeled heterogeneous target domain. Although UDA is …

Brain Tumor SegmentationDomain AdaptationImage SegmentationMedical Image Segmentation+5

Deep Residual Correction Network for Partial Domain Adaptation

2020-04-10 · Shuang Li, Chi Harold Liu, Qiuxia Lin, Qi Wen 외

Deep domain adaptation methods have achieved appealing performance by learning transferable representations from a well-labeled source domain to a different but related unlabeled target domain. Most existing works assume…

Domain AdaptationPartial Domain Adaptation