A Causal View of Entity Bias in (Large) Language Models

• URL: http://arxiv.org/abs/2305.14695v2
• Date: Thu, 26 Oct 2023 19:02:27 GMT
• Title: A Causal View of Entity Bias in (Large) Language Models
• Abstract summary: We propose causal intervention techniques to mitigate entity bias for both white-box and black-box settings. Under the white-box setting, our training-time intervention improves OOD performance of PLMs on relation extraction (RE) and machine reading comprehension (MRC) by 5.7 points and by 9.1 points, respectively. Under the black-box setting, our in-context intervention effectively reduces the entity-based knowledge conflicts of GPT-3.5, achieving up to 20.5 points of improvement of exact match accuracy on MRC and up to 17.6 points of reduction in memorization ratio on RE.
• Score: 42.64248130399508
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Entity bias widely affects pretrained (large) language models, causing them to rely on (biased) parametric knowledge to make unfaithful predictions. Although causality-inspired methods have shown great potential to mitigate entity bias, it is hard to precisely estimate the parameters of underlying causal models in practice. The rise of black-box LLMs also makes the situation even worse, because of their inaccessible parameters and uncalibrated logits. To address these problems, we propose a specific structured causal model (SCM) whose parameters are comparatively easier to estimate. Building upon this SCM, we propose causal intervention techniques to mitigate entity bias for both white-box and black-box settings. The proposed causal intervention perturbs the original entity with neighboring entities. This intervention reduces specific biasing information pertaining to the original entity while still preserving sufficient semantic information from similar entities. Under the white-box setting, our training-time intervention improves OOD performance of PLMs on relation extraction (RE) and machine reading comprehension (MRC) by 5.7 points and by 9.1 points, respectively. Under the black-box setting, our in-context intervention effectively reduces the entity-based knowledge conflicts of GPT-3.5, achieving up to 20.5 points of improvement of exact match accuracy on MRC and up to 17.6 points of reduction in memorization ratio on RE. Our code is available at https://github.com/luka-group/Causal-View-of-Entity-Bias.

Related papers

• Unbiased Alignment for Large Language Models with Noisy Preferences [74.69408994904965]
  We present a theoretical framework for unbiased alignment, introducing the Unbiased Reward Model (URM) loss and the Unbiased Direct Preference Optimization (UDPO) loss.<n>By mathematically correcting the distortion induced by preference noise, our novel objectives enable unbiased model training directly from noisy datasets.
  arXiv  Detail & Related papers  (2026-07-03T12:04:43Z)
• ADKO: Agentic Decentralized Knowledge Optimization [21.537388090124196]
  Agentic Decentralized Knowledge Optimization (ADKO) is a framework for collaborative black-box optimization across autonomous agents.<n>Each agent maintains a private Gaussian Process (GP) surrogate trained on local data and communicates only through knowledge tokens.<n>We provide the first formal analysis of dual information loss: token compression, quantified via mutual-information-based fidelity, and LM approximation, decomposed into bias and noise.
  arXiv  Detail & Related papers  (2026-05-08T15:23:53Z)
• Quantifying Prior Dominance in RAG Systems [1.9544213396776273]
  We introduce the Normalized Context Utilization (NCU) metric to quantify contextual information gain.<n>We evaluate architectures ranging from 1.5B to 72B parameters alongside a proprietary commercial API.<n>Our findings highlight the structural advantage and superior contextual adherence of SLMs in strict extraction.
  arXiv  Detail & Related papers  (2026-04-29T15:38:24Z)
• SEM: Sparse Embedding Modulation for Post-Hoc Debiasing of Vision-Language Models [36.40093869747498]
  We propose Sparse Embedding Modulation (SEM), a post-hoc, zero-shot debiasing framework that operates in a Sparse Autoencoder latent space.<n>SEM identifies and modulates bias-relevant neurons while preserving query-relevant ones.<n>Our results demonstrate that sparse latent representations provide an effective foundation for post-hoc debiasing of vision-language models.
  arXiv  Detail & Related papers  (2026-03-19T15:28:08Z)
• An Empirical Survey of Model Merging Algorithms for Social Bias Mitigation [0.9430947207126281]
  Large language models (LLMs) are known to inherit and even amplify societal biases present in their pre-training corpora.<n>We empirically survey seven algorithms: Linear, Karcher Mean, SLERP, NuSLERP, TIES, DELLA, and Nearswap, applying 13 open weight models in the GPT, LLaMA, and Qwen families.<n>We find a trade-off between bias reduction and downstream performance: methods achieving greater bias mitigation degrade accuracy.
  arXiv  Detail & Related papers  (2025-12-02T12:18:48Z)
• Bayesian Semiparametric Causal Inference: Targeted Doubly Robust Estimation of Treatment Effects [1.2833734915643464]
  We propose a semiparametric Bayesian methodology for estimating the average treatment effect (ATE) within the potential outcomes framework.<n>Our method introduces a Bayesian debiasing procedure that corrects for bias arising from nuisance estimation.<n>Extensive simulations confirm the theoretical results, demonstrating accurate point estimation and credible intervals with nominal coverage.
  arXiv  Detail & Related papers  (2025-11-19T22:15:04Z)
• Gradient Extrapolation for Debiased Representation Learning [7.183424522250937]
  Gradient Extrapolation for Debiased Representation Learning (GERNE) is designed to learn debiased representations in both known and unknown attribute training cases.<n>Our analysis shows that when the extrapolated gradient points toward the batch gradient with fewer spurious correlations, it effectively guides training toward learning a debiased model.
  arXiv  Detail & Related papers  (2025-03-17T14:48:57Z)
• Impact of Adversarial Attacks on Deep Learning Model Explainability [0.0]
  We investigate the impact of adversarial attacks on the explainability of deep learning models. Our research focuses on the robustness of these explanations when models are subjected to adversarial attacks.
  arXiv  Detail & Related papers  (2024-12-15T08:41:37Z)
• Revisiting Essential and Nonessential Settings of Evidential Deep Learning [70.82728812001807]
  Evidential Deep Learning (EDL) is an emerging method for uncertainty estimation. We propose Re-EDL, a simplified yet more effective variant of EDL.
  arXiv  Detail & Related papers  (2024-10-01T04:27:07Z)
• Surgical Feature-Space Decomposition of LLMs: Why, When and How? [8.826164604720738]
  We empirically study the efficacy of weight and feature space decomposition in transformer-based language models. We show that surgical decomposition provides critical insights into the trade-off between compression and language modelling performance. We extend our investigation to the implications of low-rank approximations on model bias.
  arXiv  Detail & Related papers  (2024-05-17T07:34:03Z)
• Dissecting Human and LLM Preferences [80.55271307662365]
  We find that humans are less sensitive to errors, favor responses that support their stances, and show clear dislike when models admit their limits. advanced LLMs like GPT-4-Turbo emphasize correctness, clarity, and harmlessness more. We show that preference-based evaluation can be intentionally manipulated.
  arXiv  Detail & Related papers  (2024-02-17T14:34:31Z)
• Marginal Debiased Network for Fair Visual Recognition [59.05212866862219]
  We propose a novel marginal debiased network (MDN) to learn debiased representations. Our MDN can achieve a remarkable performance on under-represented samples.
  arXiv  Detail & Related papers  (2024-01-04T08:57:09Z)
• Impact of Co-occurrence on Factual Knowledge of Large Language Models [25.89517624732426]
  Large language models (LLMs) often make factually incorrect responses despite their success in various applications. Our results reveal that LLMs are vulnerable to the co-occurrence bias, defined as preferring frequently co-occurred words over the correct answer.
  arXiv  Detail & Related papers  (2023-10-12T12:01:32Z)
• Robust Natural Language Understanding with Residual Attention Debiasing [28.53546504339952]
  We propose an end-to-end debiasing method that mitigates unintended biases from attention. Experiments show that READ significantly improves the performance of BERT-based models on OOD data with shortcuts removed.
  arXiv  Detail & Related papers  (2023-05-28T04:25:04Z)
• ADEPT: A DEbiasing PrompT Framework [49.582497203415855]
  Finetuning is an applicable approach for debiasing contextualized word embeddings. discrete prompts with semantic meanings have shown to be effective in debiasing tasks. We propose ADEPT, a method to debias PLMs using prompt tuning while maintaining the delicate balance between removing biases and ensuring representation ability.
  arXiv  Detail & Related papers  (2022-11-10T08:41:40Z)
• Unifying Model Explainability and Robustness for Joint Text Classification and Rationale Extraction [11.878012909876713]
  We propose a joint classification and rationale extraction model named AT-BMC. It includes two key mechanisms: mixed Adversarial Training (AT) is designed to use various perturbations in discrete and embedding space to improve the model's robustness, and Boundary Match Constraint (BMC) helps to locate rationales more precisely with the guidance of boundary information. Performances on benchmark datasets demonstrate that the proposed AT-BMC outperforms baselines on both classification and rationale extraction by a large margin.
  arXiv  Detail & Related papers  (2021-12-20T09:48:32Z)
• Debiased Explainable Pairwise Ranking from Implicit Feedback [0.3867363075280543]
  We focus on the state of the art pairwise ranking model, Bayesian Personalized Ranking (BPR) BPR is a black box model that does not explain its outputs, thus limiting the user's trust in the recommendations. We propose a novel explainable loss function and a corresponding Matrix Factorization-based model that generates recommendations along with item-based explanations.
  arXiv  Detail & Related papers  (2021-07-30T17:19:37Z)
• Deconfounding Scores: Feature Representations for Causal Effect Estimation with Weak Overlap [140.98628848491146]
  We introduce deconfounding scores, which induce better overlap without biasing the target of estimation. We show that deconfounding scores satisfy a zero-covariance condition that is identifiable in observed data. In particular, we show that this technique could be an attractive alternative to standard regularizations.
  arXiv  Detail & Related papers  (2021-04-12T18:50:11Z)

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.