Bias Leaves a Gradient Trail: Label-Free Bias Identification via Gradient Probes on Concept Decompositions

• URL: http://arxiv.org/abs/2605.28780v1
• Date: Wed, 27 May 2026 17:39:10 GMT
• Title: Bias Leaves a Gradient Trail: Label-Free Bias Identification via Gradient Probes on Concept Decompositions
• Abstract summary: We present a bias-label-free, post-hoc method for identifying spurious concepts in frozen vision models.<n>On Colored MNIST and Waterbirds the method recovers concepts aligned with the known spurious cue.<n> suppressing the top-ranked concepts at inference time improves worst-group accuracy by up to 17.9 percentage points.
• Score: 8.894428198928333
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Vision classifiers can exploit spurious correlations, achieving high in-distribution accuracy yet failing under distribution shift. Existing approaches to bias mitigation and analysis often depend on curated datasets, spurious-attribute or group labels, or retraining, which may be infeasible once a model is deployed or the relevant bias is unknown. We present a bias-label-free, post-hoc method for identifying spurious concepts in frozen vision models, relying only on standard class labels from a held-out audit dataset. For each target class, we collect patches from inputs predicted as that class and apply non-negative matrix factorization to intermediate activations to obtain a bank of interpretable concept vectors. Candidate concepts are then ranked with a bias estimator derived from their interaction with backpropagated gradients on misclassified examples: bias concepts tend to get activated when correcting false negatives and suppressed when correcting false positives. On Colored MNIST and Waterbirds the method recovers concepts aligned with the known spurious cue, and on CelebA it surfaces decision-relevant directions that only partially coincide with the annotated gender attribute; suppressing the top-ranked concepts at inference time improves worst-group accuracy by up to 17.9 percentage points on Waterbirds and 10.4 on CelebA without any retraining or parameter updates. Our method identifies decision-relevant spurious directions that need not coincide with annotated ones, providing both an interpretable auditing tool and an actionable debiasing handle for frozen vision models. Code is available at https://github.com/vitryt/label-free-bias-identification.

Related papers

• Dual-Branch Cross-Projection Debiasing through Diffusion-based Disentanglement [66.67463557274358]
  Foundation models trained on biased datasets often rely on spurious correlations between target labels and non-causal attributes.<n>We introduce Confidence-guided Concept Mining (CBCM) to identify reliable spurious attributes without attribute annotations.<n>We propose Dual-branch Cross-projection Debiasing (DCD), a prompt-tuning framework that separates target and spurious representations into two branches.
  arXiv  Detail & Related papers  (2026-06-23T05:28:47Z)
• Towards Fairness under Label Bias in Image Segmentation: Impact, Measurement and Mitigation [10.15221228043609]
  Labeled datasets reflect the biases of their annotation pipelines, which sometimes introduce label bias.<n>We present a data-centric adaptation of Confident Learning to segmentation, allowing detection of label bias directly in the training data.<n>We show how our framework reliably detects and mitigates bias without access to clean labels, achieving equitable performance across experimental conditions.
  arXiv  Detail & Related papers  (2026-05-07T19:47:50Z)
• Learning from Emptiness: De-biasing Listwise Rerankers with Content-Agnostic Probability Calibration [76.08899010904652]
  CapCal is a training-free framework that mechanically decouples positional bias from ranking decisions.<n>It achieves superior performance among training-free methods while preserving single-pass efficiency.
  arXiv  Detail & Related papers  (2026-04-11T10:47:22Z)
• Reference-Free Rating of LLM Responses via Latent Information [53.463883683503106]
  We study the common practice of asking a judge model to assign Likert-scale scores to free-text responses.<n>We then propose and evaluate Latent Judges, which derive scalar ratings from internal model signals.<n>Across a broad suite of pairwise and single-rating benchmarks, latent methods match or surpass standard prompting.
  arXiv  Detail & Related papers  (2025-09-29T12:15:52Z)
• ShortcutProbe: Probing Prediction Shortcuts for Learning Robust Models [26.544938760265136]
  Deep learning models inadvertently learn spurious correlations between targets and non-essential features.<n>In this paper, we propose a novel post hoc spurious bias mitigation framework without requiring group labels.<n>Our framework, termed ShortcutProbe, identifies prediction shortcuts that reflect potential non-robustness in predictions in a given model's latent space.
  arXiv  Detail & Related papers  (2025-05-20T04:21:17Z)
• CUBIC: Concept Embeddings for Unsupervised Bias Identification using VLMs [2.0062715282793233]
  Methods that interpret high, human-understandable concepts are more effective than those relying on low-level features like heat.<n>A major challenge for concept-based methods is the lack of image annotations indicating such labeling concepts.<n>We present CUBIC, which does not rely on predefined candidates or examples of model failures tied to specific biases.
  arXiv  Detail & Related papers  (2025-05-16T09:57:15Z)
• Efficient Online Set-valued Classification with Bandit Feedback [10.882001129426726]
  We propose Bandit Class-specific Conformal Prediction (BCCP), offering coverage guarantees on a class-specific granularity. BCCP overcomes the challenges of sparsely labeled data in each iteration and generalizes the reliability and applicability of conformal prediction to online decision-making environments.
  arXiv  Detail & Related papers  (2024-05-07T15:14:51Z)
• Dist-PU: Positive-Unlabeled Learning from a Label Distribution Perspective [89.5370481649529]
  We propose a label distribution perspective for PU learning in this paper. Motivated by this, we propose to pursue the label distribution consistency between predicted and ground-truth label distributions. Experiments on three benchmark datasets validate the effectiveness of the proposed method.
  arXiv  Detail & Related papers  (2022-12-06T07:38:29Z)
• Self-supervised debiasing using low rank regularization [59.84695042540525]
  Spurious correlations can cause strong biases in deep neural networks, impairing generalization ability. We propose a self-supervised debiasing framework potentially compatible with unlabeled samples. Remarkably, the proposed debiasing framework significantly improves the generalization performance of self-supervised learning baselines.
  arXiv  Detail & Related papers  (2022-10-11T08:26:19Z)
• Disentangling Sampling and Labeling Bias for Learning in Large-Output Spaces [64.23172847182109]
  We show that different negative sampling schemes implicitly trade-off performance on dominant versus rare labels. We provide a unified means to explicitly tackle both sampling bias, arising from working with a subset of all labels, and labeling bias, which is inherent to the data due to label imbalance.
  arXiv  Detail & Related papers  (2021-05-12T15:40:13Z)
• Exploiting Sample Uncertainty for Domain Adaptive Person Re-Identification [137.9939571408506]
  We estimate and exploit the credibility of the assigned pseudo-label of each sample to alleviate the influence of noisy labels. Our uncertainty-guided optimization brings significant improvement and achieves the state-of-the-art performance on benchmark datasets.
  arXiv  Detail & Related papers  (2020-12-16T04:09:04Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.