Papers
arxiv:2611.22668

Let Samples Sing in Chorus: Mitigating Spurious Correlations via Cross-view Consensus

Published on Nov 17
· Submitted by
Dave Lee
on Nov 19
Authors:
,
,
,
,

Abstract

A multi-view debiasing method aggregates within-class partitions across diverse representation spaces to reduce reliance on spurious correlations and improve worst-group accuracy without bias supervision.

Deep models often rely on spurious correlations that fail on underrepresented groups. Existing methods either correct human-specified bias attributes or infer unknown biases from the learned representation space. However, annotated attributes may differ from the shortcuts actually used by a model, while a single representation provides incomplete and potentially noisy evidence. Our key insight is that a bias obscured in one feature space may be clearer in another. In this work, we propose CHORD (Cross-view Harmonic Outlier Reversal for Debiasing), a multi-view framework that allows samples to "sing in chorus" across diverse representation spaces to achieve robust cross-view consensus. CHORD aggregates aggregates within-class partitions in each view, which is a proxy for specifying candidate biases, to debias the target representation and classifier without bias supervision. Experiments on seven benchmarks across vision, language, speech, and medical imaging show consistent improvements in worst-group accuracy. Code and checkpoints are available at https://github.com/davelee-uestc/chord_debiasing_public .

Community

Paper author Paper submitter

We proposed CHORD (Cross-view Harmonic Outlier Reversal for Debiasing), a multi-view framework that allows samples to "sing in chorus" across diverse representation spaces to achieve robust cross-view consensus.

Hi authors, thanks for the work and the released checkpoints. A quick question: the CelebA target-only WGA is 84.27 in Tables 4 and 5 but 82.39 in the Discussion, with the same ERM model, 10 classifier seeds and CHORD at 89.06 in both; 84.27 is also the NSF baseline value in Table 1. Details in this GitHub issue: https://github.com/davelee-uestc/chord_debiasing_public/issues/2

·

Thanks for your question. 82.39% is the target-only result from our released checkpoint, used in the paired Discussion experiment. In Tables 4–5, we use the higher, previously reported NSF result of 84.27% to avoid overstating CHORD’s improvement. The two values therefore correspond to different checkpoint settings. We provide checkpoints for all our experimental results to support reproducibility.

This is an automated message from the Librarian Bot. I found the following papers similar to this paper.

The following papers were recommended by the Semantic Scholar API

Please give a thumbs up to this comment if you found it helpful!

If you want recommendations for any Paper on Hugging Face checkout this Space

You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: @librarian-bot recommend

Sign up or log in to comment

Get this paper in your agent:

hf papers read 2611.22668
Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash

Models citing this paper 0

No model linking this paper

Cite arxiv.org/abs/2611.22668 in a model README.md to link it from this page.

Datasets citing this paper 0

No dataset linking this paper

Cite arxiv.org/abs/2611.22668 in a dataset README.md to link it from this page.

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2611.22668 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.