Context-aware Fairness Evaluation and Mitigation in LLMs

• URL: http://arxiv.org/abs/2510.18914v1
• Date: Tue, 21 Oct 2025 06:13:17 GMT
• Title: Context-aware Fairness Evaluation and Mitigation in LLMs
• Authors: Afrozah Nadeem, Mark Dras, Usman Naseem,
• Abstract summary: Large language models often display undesirable behaviors embedded in their internal representations.<n>Pruning-based methods provide a flexible and transparent way to reduce bias by adjusting the neurons responsible for certain behaviors.<n>We propose a dynamic, reversible, pruning-based framework that detects context-aware neuron activations and applies adaptive masking to modulate their influence during generation.
• Score: 16.594400974742523
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models often display undesirable behaviors embedded in their internal representations, undermining fairness, inconsistency drift, amplification of harmful content, and the propagation of unwanted patterns during extended dialogue and conversations. Although training-time or data-centric methods attempt to reduce these effects, they are computationally expensive, irreversible once deployed, and slow to adapt to new conversational contexts. Pruning-based methods provide a flexible and transparent way to reduce bias by adjusting the neurons responsible for certain behaviors. However, most existing approaches are static; once a neuron is removed, the model loses the ability to adapt when the conversation or context changes. To address this, we propose a dynamic, reversible, pruning-based framework that detects context-aware neuron activations and applies adaptive masking to modulate their influence during generation. Our inference-time solution provides fine-grained, memory-aware mitigation with knowledge-preserved, more coherent behavior across multilingual single- and multi-turn dialogues, enabling dynamic fairness control in real-world conversational AI.

Related papers

• InteracTalker: Prompt-Based Human-Object Interaction with Co-Speech Gesture Generation [1.7523719472700858]
  We introduce InteracTalker, a novel framework that seamlessly integrates prompt-based object-aware interactions with co-speech gesture generation.<n>Our framework utilizes a Generalized Motion Adaptation Module that enables independent training, adapting to the corresponding motion condition.<n>InteracTalker successfully unifies these previously separate tasks, outperforming prior methods in both co-speech gesture generation and object-interaction synthesis.
  arXiv  Detail & Related papers  (2025-12-14T12:29:49Z)
• Chronological Thinking in Full-Duplex Spoken Dialogue Language Models [66.84843878538207]
  Chronological Thinking aims to improve response quality in full SDLMs.<n>No additional latency: once the user stops speaking, the agent halts thinking and begins speaking without further delay.<n>Results: Experiments demonstrate the effectiveness of chronological thinking through both objective metrics and human evaluations.
  arXiv  Detail & Related papers  (2025-10-02T10:28:11Z)
• Explaining multimodal LLMs via intra-modal token interactions [55.27436637894534]
  Multimodal Large Language Models (MLLMs) have achieved remarkable success across diverse vision-language tasks, yet their internal decision-making mechanisms remain insufficiently understood.<n>We propose enhancing interpretability by leveraging intra-modal interaction.
  arXiv  Detail & Related papers  (2025-09-26T14:39:13Z)
• Mitigating Harmful Erraticism in LLMs Through Dialectical Behavior Therapy Based De-Escalation Strategies [0.0]
  This paper hypothesizes that a framework rooted in human psychological principles, specifically therapeutic modalities, can provide a more robust and sustainable solution.<n> Drawing an analogy to the simulated neural networks of AI mirroring the human brain, we propose the application of Dialectical Behavior Therapy (DBT) principles.
  arXiv  Detail & Related papers  (2025-09-06T11:20:15Z)
• Aligning Spoken Dialogue Models from User Interactions [55.192134724622235]
  We propose a novel preference alignment framework to improve spoken dialogue models on realtime conversations from user interactions.<n>We create a dataset of more than 150,000 preference pairs from raw multi-turn speech conversations annotated with AI feedback.<n>Our findings shed light on the importance of a well-calibrated balance among various dynamics, crucial for natural real-time speech dialogue systems.
  arXiv  Detail & Related papers  (2025-06-26T16:45:20Z)
• Effective Context Modeling Framework for Emotion Recognition in Conversations [2.7175580940471913]
  Emotion Recognition in Conversations (ERC) facilitates a deeper understanding of the emotions conveyed by speakers in each utterance within a conversation.<n>Recent Graph Neural Networks (GNNs) have demonstrated their strengths in capturing data relationships.<n>We propose ConxGNN, a novel GNN-based framework designed to capture contextual information in conversations.
  arXiv  Detail & Related papers  (2024-12-21T02:22:06Z)
• Multi-level Adaptive Contrastive Learning for Knowledge Internalization in Dialogue Generation [37.55417272177113]
  Knowledge-grounded dialogue generation aims to incorporate external knowledge to supplement the context. However, the model often fails to internalize this information into responses in a human-like manner. We propose a Multi-level Adaptive Contrastive Learning framework that dynamically samples negative examples and subsequently penalizes degeneration behaviors.
  arXiv  Detail & Related papers  (2023-10-13T08:16:27Z)
• Color Overmodification Emerges from Data-Driven Learning and Pragmatic Reasoning [53.088796874029974]
  We show that speakers' referential expressions depart from communicative ideals in ways that help illuminate the nature of pragmatic language use. By adopting neural networks as learning agents, we show that overmodification is more likely with environmental features that are infrequent or salient.
  arXiv  Detail & Related papers  (2022-05-18T18:42:43Z)
• Learning Adaptive Language Interfaces through Decomposition [89.21937539950966]
  We introduce a neural semantic parsing system that learns new high-level abstractions through decomposition. Users interactively teach the system by breaking down high-level utterances describing novel behavior into low-level steps.
  arXiv  Detail & Related papers  (2020-10-11T08:27:07Z)
• A Controllable Model of Grounded Response Generation [122.7121624884747]
  Current end-to-end neural conversation models inherently lack the flexibility to impose semantic control in the response generation process. We propose a framework that we call controllable grounded response generation (CGRG) We show that using this framework, a transformer based model with a novel inductive attention mechanism, trained on a conversation-like Reddit dataset, outperforms strong generation baselines.
  arXiv  Detail & Related papers  (2020-05-01T21:22:08Z)

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.