IR-guided diffusion injects intermediate text representations into early denoising steps to improve alignment for one-and-only objects, reporting up to 19.1pp VQAScore gains on OAO-AttackBench and other benchmarks.
Debiasing vision-language models via biased prompts
9 Pith papers cite this work. Polarity classification is still indexing.
9
Pith papers citing it
representative citing papers
DAS adds a cached per-class bonus derived from noise-conditioned text prompts to cosine scores, improving accuracy by 2.60-5.75 points on UrbanSound8K and mAP by 1.50-1.74 points on FSD50K under urban noise.
DAT rescales CLIP image-text similarities based on local embedding density to reduce the impact of spurious correlations in zero-shot classification.
StayFair addresses guidance bias in diffusion models by extending demographic parity, allowing fairness to hold across guidance scales via modified classifier or null-embedding steps.
Embedding Arithmetic performs vector operations in the embedding space of T2I models to mitigate bias at inference time, outperforming baselines on diversity while preserving coherence via a new Concept Coherence Score.
TES applies early global alignment then iterative CLIP-guided refinement to text embeddings in Stable Diffusion to mitigate bias while preserving quality.
Zero-shot vision-language models are unreliable and vary widely for depression screening, and explainability-based fairness interventions often trade away accuracy without reliable fairness gains.
POVID generates AI-created preference data to fine-tune vision-language models with DPO, reducing hallucinations and improving benchmark scores.
A systematic review of T2I bias literature that distinguishes target and threshold fairness and proposes a target-based operationalization framework.
citing papers explorer
-
Intermediate Text Representation Guided Text-to-Image Generation for Enhancing One-and-Only Alignment
IR-guided diffusion injects intermediate text representations into early denoising steps to improve alignment for one-and-only objects, reporting up to 19.1pp VQAScore gains on OAO-AttackBench and other benchmarks.
-
Drift-Augmented Scoring: Text-Derived Noise Robustness for Zero-Shot Audio-Language Classification
DAS adds a cached per-class bonus derived from noise-conditioned text prompts to cosine scores, improving accuracy by 2.60-5.75 points on UrbanSound8K and mAP by 1.50-1.74 points on FSD50K under urban noise.
-
Density-Aware Translation of Spurious Correlations in Zero-Shot VLMs
DAT rescales CLIP image-text similarities based on local embedding density to reduce the impact of spurious correlations in zero-shot classification.
-
Stay Fair! Ensuring Group Fairness in Diffusion Models Across Guidance Scales
StayFair addresses guidance bias in diffusion models by extending demographic parity, allowing fairness to hold across guidance scales via modified classifier or null-embedding steps.
-
Embedding Arithmetic: A Lightweight, Tuning-Free Framework for Post-hoc Bias Mitigation in Text-to-Image Models
Embedding Arithmetic performs vector operations in the embedding space of T2I models to mitigate bias at inference time, outperforming baselines on diversity while preserving coherence via a new Concept Coherence Score.
-
Training-Free Debiasing of Diffusion Models via CLIP-Guided Denoising Optimization
TES applies early global alignment then iterative CLIP-guided refinement to text embeddings in Stable Diffusion to mitigate bias while preserving quality.
-
FAIR_XAI: Improving Multimodal Foundation Model Fairness via Explainability for Wellbeing Assessment
Zero-shot vision-language models are unreliable and vary widely for depression screening, and explainability-based fairness interventions often trade away accuracy without reliable fairness gains.
-
Aligning Modalities in Vision Large Language Models via Preference Fine-tuning
POVID generates AI-created preference data to fine-tune vision-language models with DPO, reducing hallucinations and improving benchmark scores.
-
Operationalizing Fairness in Text-to-Image Models: A Survey of Bias, Fairness Audits and Mitigation Strategies
A systematic review of T2I bias literature that distinguishes target and threshold fairness and proposes a target-based operationalization framework.