Diagnosing the Performance Trade-off in Moral Alignment: A Case Study on Gender Stereotypes

• URL: http://arxiv.org/abs/2509.21456v2
• Date: Tue, 07 Oct 2025 16:39:27 GMT
• Title: Diagnosing the Performance Trade-off in Moral Alignment: A Case Study on Gender Stereotypes
• Abstract summary: Moral alignment has emerged as a widely adopted approach for regulating the behavior of pretrained language models.<n>Gender stereotype mitigation is a representational task within the broader application of moral alignment.<n>Prior studies commonly aim to achieve a performance trade-off by encouraging PLMs to selectively forget only stereotypical knowledge.
• Score: 9.264601359430799
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Moral alignment has emerged as a widely adopted approach for regulating the behavior of pretrained language models (PLMs), typically through fine-tuning on curated datasets. Gender stereotype mitigation is a representational task within the broader application of moral alignment. However, this process often comes at the cost of degraded downstream task performance. Prior studies commonly aim to achieve a performance trade-off by encouraging PLMs to selectively forget only stereotypical knowledge through carefully designed fairness objective, while preserving their language modeling capability (overall forgetting). In this short paper, we investigate whether the performance trade-off can be achieved through the lens of forgetting and the fairness objective. Our analysis shows that the large datasets needed for satisfactory fairness highlight the limitations of current fairness objectives in achieving an effective trade-off: (1) downstream task performance is strongly correlated with overall forgetting; (2) selective forgetting reduces stereotypes, but overall forgetting increases. and (3) general solutions for alleviating forgetting are ineffective at reducing the overall forgetting and fail to improve downstream task performance.

Related papers

• Can Vision Language Models Judge Action Quality? An Empirical Evaluation [5.688409551177917]
  Action Quality Assessment (AQA) has broad applications in physical therapy, sports coaching, and competitive judging.<n>Although Vision Language Models (VLMs) hold considerable promise for AQA, their actual performance in this domain remains largely uncharacterised.<n>We present a comprehensive evaluation of state-of-the-art VLMs across activity domains (e.g. fitness, figure skating, diving), tasks, representations, and prompting strategies.
  arXiv  Detail & Related papers  (2026-04-09T14:29:19Z)
• IntentionVLA: Generalizable and Efficient Embodied Intention Reasoning for Human-Robot Interaction [51.130510883952546]
  Vision-Language-Action (VLA) models leverage pretrained vision-language models (VLMs) to couple perception with robotic control.<n>We propose textbfIntentionVLA, a VLA framework with a curriculum training paradigm and an efficient inference mechanism.<n>Our proposed method first leverages carefully designed reasoning data that combine intention inference, spatial grounding, and compact embodied reasoning.
  arXiv  Detail & Related papers  (2025-10-09T04:49:46Z)
• Intrinsic Meets Extrinsic Fairness: Assessing the Downstream Impact of Bias Mitigation in Large Language Models [11.396244643030983]
  Large Language Models (LLMs) exhibit socio-economic biases that can propagate into downstream tasks.<n>We present a unified evaluation framework to compare intrinsic bias mitigation via concept unlearning with extrinsic bias mitigation via counterfactual data augmentation.<n>Our results show that intrinsic bias mitigation through unlearning reduces intrinsic gender bias by up to 94.9%, while also improving downstream task fairness metrics, such as demographic parity by up to 82%, without compromising accuracy.
  arXiv  Detail & Related papers  (2025-09-19T22:59:55Z)
• Fair Deepfake Detectors Can Generalize [51.21167546843708]
  We show that controlling for confounders (data distribution and model capacity) enables improved generalization via fairness interventions.<n>Motivated by this insight, we propose Demographic Attribute-insensitive Intervention Detection (DAID), a plug-and-play framework composed of: i) Demographic-aware data rebalancing, which employs inverse-propensity weighting and subgroup-wise feature normalization to neutralize distributional biases; and ii) Demographic-agnostic feature aggregation, which uses a novel alignment loss to suppress sensitive-attribute signals.<n>DAID consistently achieves superior performance in both fairness and generalization compared to several state-of-the-art
  arXiv  Detail & Related papers  (2025-07-03T14:10:02Z)
• Towards Harmless Rawlsian Fairness Regardless of Demographic Prior [57.30787578956235]
  We explore the potential for achieving fairness without compromising its utility when no prior demographics are provided to the training set. We propose a simple but effective method named VFair to minimize the variance of training losses inside the optimal set of empirical losses.
  arXiv  Detail & Related papers  (2024-11-04T12:40:34Z)
• Identifying and Mitigating Social Bias Knowledge in Language Models [52.52955281662332]
  We propose a novel debiasing approach, Fairness Stamp (FAST), which enables fine-grained calibration of individual social biases.<n>FAST surpasses state-of-the-art baselines with superior debiasing performance.<n>This highlights the potential of fine-grained debiasing strategies to achieve fairness in large language models.
  arXiv  Detail & Related papers  (2024-08-07T17:14:58Z)
• Enhancing Robustness of Vision-Language Models through Orthogonality Learning and Self-Regularization [77.62516752323207]
  We introduce an orthogonal fine-tuning method for efficiently fine-tuning pretrained weights and enabling enhanced robustness and generalization. A self-regularization strategy is further exploited to maintain the stability in terms of zero-shot generalization of VLMs, dubbed OrthSR. For the first time, we revisit the CLIP and CoOp with our method to effectively improve the model on few-shot image classficiation scenario.
  arXiv  Detail & Related papers  (2024-07-11T10:35:53Z)
• Low-rank finetuning for LLMs: A fairness perspective [54.13240282850982]
  Low-rank approximation techniques have become the de facto standard for fine-tuning Large Language Models. This paper investigates the effectiveness of these methods in capturing the shift of fine-tuning datasets from the initial pre-trained data distribution. We show that low-rank fine-tuning inadvertently preserves undesirable biases and toxic behaviors.
  arXiv  Detail & Related papers  (2024-05-28T20:43:53Z)
• Augmenting Unsupervised Reinforcement Learning with Self-Reference [63.68018737038331]
  Humans possess the ability to draw on past experiences explicitly when learning new tasks. We propose the Self-Reference (SR) approach, an add-on module explicitly designed to leverage historical information. Our approach achieves state-of-the-art results in terms of Interquartile Mean (IQM) performance and Optimality Gap reduction on the Unsupervised Reinforcement Learning Benchmark.
  arXiv  Detail & Related papers  (2023-11-16T09:07:34Z)
• Continual Learners are Incremental Model Generalizers [70.34479702177988]
  This paper extensively studies the impact of Continual Learning (CL) models as pre-trainers. We find that the transfer quality of the representation often increases gradually without noticeable degradation in fine-tuning performance. We propose a new fine-tuning scheme, GLobal Attention Discretization (GLAD), that preserves rich task-generic representation during solving downstream tasks.
  arXiv  Detail & Related papers  (2023-06-21T05:26:28Z)
• Model Debiasing via Gradient-based Explanation on Representation [14.673988027271388]
  We propose a novel fairness framework that performs debiasing with regard to sensitive attributes and proxy attributes. Our framework achieves better fairness-accuracy trade-off on unstructured and structured datasets than previous state-of-the-art approaches.
  arXiv  Detail & Related papers  (2023-05-20T11:57:57Z)
• Fairness and Explainability: Bridging the Gap Towards Fair Model Explanations [12.248793742165278]
  We bridge the gap between fairness and explainability by presenting a novel perspective of procedure-oriented fairness based on explanations. We propose a Comprehensive Fairness Algorithm (CFA), which simultaneously fulfills multiple objectives - improving traditional fairness, satisfying explanation fairness, and maintaining the utility performance.
  arXiv  Detail & Related papers  (2022-12-07T18:35:54Z)
• Self-Supervised Learning via Maximum Entropy Coding [57.56570417545023]
  We propose Maximum Entropy Coding (MEC) as a principled objective that explicitly optimize on the structure of the representation. MEC learns a more generalizable representation than previous methods based on specific pretext tasks. It achieves state-of-the-art performance consistently on various downstream tasks, including not only ImageNet linear probe, but also semi-supervised classification, object detection, instance segmentation, and object tracking.
  arXiv  Detail & Related papers  (2022-10-20T17:58:30Z)
• Optimising Equal Opportunity Fairness in Model Training [60.0947291284978]
  Existing debiasing methods, such as adversarial training and removing protected information from representations, have been shown to reduce bias. We propose two novel training objectives which directly optimise for the widely-used criterion of it equal opportunity, and show that they are effective in reducing bias while maintaining high performance over two classification tasks.
  arXiv  Detail & Related papers  (2022-05-05T01:57:58Z)
• Supercharging Imbalanced Data Learning With Energy-based Contrastive Representation Transfer [72.5190560787569]
  In computer vision, learning from long tailed datasets is a recurring theme, especially for natural image datasets. Our proposal posits a meta-distributional scenario, where the data generating mechanism is invariant across the label-conditional feature distributions. This allows us to leverage a causal data inflation procedure to enlarge the representation of minority classes.
  arXiv  Detail & Related papers  (2020-11-25T00:13:11Z)
• Ethical Adversaries: Towards Mitigating Unfairness with Adversarial Machine Learning [8.436127109155008]
  Individuals, as well as organisations, notice, test, and criticize unfair results to hold model designers and deployers accountable. We offer a framework that assists these groups in mitigating unfair representations stemming from the training datasets. Our framework relies on two inter-operating adversaries to improve fairness.
  arXiv  Detail & Related papers  (2020-05-14T10:10:19Z)

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.