2026· Annual Meeting of the Association for Computational Linguistics· pp. 27525-27539· 0 citations· 25 references
Computer Science
TL;DR
This paper introduces a prototype-guided modeling approach that leverages natural language prompts to represent confounders, transforming abstract biases into interpretable geometric anchors without auxiliary classifiers, a novel framework that mitigates spurious correlations by manipulating latent space geometry.
Abstract
Spurious correlations cause deep learning models to rely on predictive shortcuts that hold in the training data but break under distribution shifts, leading to large performance drops for minority groups. Existing strategies often rely on costly group annotations or employ unstable adversarial training. In this paper, we pro-pose Prototype-guided debiasing using Robust Invariant Feature Transformations (PRIFT), a novel framework that mitigates spurious correlations by manipulating latent space geometry. Specifically, we introduce a prototype-guided modeling approach that leverages natural language prompts to represent confounders, transforming abstract biases into interpretable geometric anchors without auxiliary classifiers. Based on these anchors, we introduce a centered projection operator that adaptively puri-fies representations by removing confounding deviations specific to instances while preserving essential semantic structure. Furthermore, PRIFT can handle confounding factor information at different levels, ranging from true labels to unsupervised latent inference. Experiments on four text classification benchmarks demonstrate the superiority of our method; notably, PRIFT outperforms state-of-the-art baselines and improves worst-group accuracy by over 20% on the CivilComments dataset compared to standard empirical risk minimization.
Spurious correlations pose a significant challenge to the robustness of modern machine learning. The inherent imbalance in dataset distributions often leads traditional Empirical Risk Minimization (ERM) models to rely on majority spurious attributes for classification, resulting in poor performance on minority groups. This problem becomes particularly challenging when the spurious attributes are unavailable. Existing group-label-free methods often upsample minority groups or misclassified real training examples; repeating the same instances can reduce effective diversity and encourage overfitting. To mitigate these spurious correlations from a data-centric perspective in the absence of prior knowledge, we introduce Subpopulation-Aware Generative Enhancement (SAGE), a two-stage generative augmentation framework. Using cluster-derived sub-labels and class labels, we fine-tune a conditional generative model and text encoder, generating targeted synthetic data to fill underrepresented regions in the training set and construct a balanced validation set for last-layer reweighting. We experimentally show that SAGE achieves 89.5%, 85.7%, and 79.1% worst-group accuracy on Waterbirds, CelebA, and MetaShift, respectively, outperforming the best group-label-free baselines by up to 7.7 percentage points.
U N M ASK is presented, a fully automated pipeline that discovers, causally verifies, and mitigates spurious correlations in text classifiers without additional human annotation, and demonstrates that the discovery and validation stages generalize to reward model preference data.
It is suggested that predictive performance, attribution plausibility, and mechanistic faithfulness characterize different aspects of model behavior and should be evaluated separately when studying explainability in media bias detection.
Tinghao Chen, Raina Zhang, Benjamin M. Ampel et al.· arXiv.org· 0 citations
Attribute correlations in the training data will compromise the ability of a deep generative model (DGM) to synthesize images with under-represented attribute combinations (i.e., minority samples). Existing approaches mitigate this by data re-sampling to remove attribute correlations seen by the DGM, using a classifier to provide pseudo-supervision on generated counterfactual samples, or incorporating inductive bias to explicitly decompose the generation into independent submechanisms. We present ProReGen, a progressive residual generation approach inspired by the classical Robinson’s transformation, to partial out from an image attribute x2 its component mx1 that is predictable by other image attributes x1, and the residual γ=x2-mx1 that is not. This simplifies the problem of learning a DGM gx1,x2 conditioned on correlated inputs, to learning g~x1,γ conditioned on orthogonal inputs. It further allows us to progressively learn g~ by first shifting the burden to abundant majority samples to learn g~x1,γ=0, and then expanding it with additional layers gres to resolve its difference to g~x1,γ using residual attribute γ on limited minority samples. On three benchmark datasets with varying strengths of attribute correlation and one dataset with natural attribute correlation, we demonstrate that ProReGen—with input orthogonalization and progressive residual learning—improved the correctness of minority generations compared to existing strategies.
Ruby Shrestha, Ajay Gopi, Casey Meisenzahl et al.· International Conference on...· 0 citations
Matching users to interest categories at scale is central to personalized shopping, but the task is challenging in large e-commerce platforms, where label spaces continually evolve and user-interest signals are sparse and long-tailed. Autoregressive language models are appealing because their world knowledge and semantic priors over descriptors generalize across extreme label spaces and accommodate multiple valid label assignments. Yet under teacher-forced fine-tuning, inference-time predictions become part of the conditioning context: early errors steer later outputs toward co-occurring labels, over-generating near-correlates and missing unrelated true interests. We present MERIT, a framework for user-interest propensity modeling that mitigates this exposure bias through a self-correction objective. A permutation-invariant multi-target loss over shuffled mixtures of gold and mined hard-negative labels exposes the generator to erroneous prefixes while preserving the efficiency of teacher-forced training. This training objective concentrates supervision at classification positions, yielding propensity-aligned hidden states powering a lightweight scorer for bidirectional retrieval (interests for users and users for interests). On a proprietary e-commerce dataset with 250k+ interest categories, MERIT improves global recall by at least 11.9% and average Hit@k by 6.1%. In production A/B tests, it achieves +0.26% gain in user conversion.
Abhinav Mahajan, Arindam Sarkar, P. Comar· 0 citations
Ordinal regression is widely used in scenarios where labels are discrete yet inherently ordered. In practice, however, ordinal labels are often obtained by discretizing underlying continuous semantics through subjective human judgment, resulting in ambiguous boundaries and annotation noise. Such uncertainty challenges existing methods that rely on fixed supervision targets, which may reinforce biased ordering under subjective annotations. To address this limitation, we propose D3O, a dynamic distribution distillation framework that replaces static supervision with training-driven evolution of ordinal label distributions via self-distillation. Specifically, we introduce a contrastive ordinal-aware label enhancement module that leverages vision-language alignment to recover refined label distributions capturing both inter-class ambiguity and instance-level uncertainty. Furthermore, we design a CDF-based cross-layer interaction distillation mechanism to propagate cumulative ordinal structure across network hierarchy, ensuring consistent ordinal geometry in intermediate representations. Extensive experiments on four general ordinal regression tasks demonstrate that our proposed D3O consistently outperforms existing approaches, particularly under severe class imbalance and noisy supervision. These results highlight the effectiveness of dynamic supervision in learning robust ordinal representations beyond fixed targets. The code will be publicly available.
