# CLIP the Bias: How Useful is Balancing Data in Multimodal Learning?

arXiv: https://arxiv.org/abs/2403.04547 · HTML: https://mlsanity.com/paper/2403.04547/

## Key facts

- Cited by 35 papers in the map (data through 14 Oct 2026).
- Builds on 60 of its 113 references that are in the map.
- First posted on arXiv 7 Mar 2024.

## Builds on (most cited first, 30 of 60)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Hierarchical Text-Conditional Image Generation with CLIP Latents](https://mlsanity.com/paper/2204.06125/) (2022)
- [Flamingo: a Visual Language Model for Few-Shot Learning](https://mlsanity.com/paper/2204.14198/) (2022)
- [Zero-Shot Text-to-Image Generation](https://mlsanity.com/paper/2102.12092/) (2021)
- [Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision](https://mlsanity.com/paper/2102.05918/) (2021)
- [Sigmoid Loss for Language Image Pre-Training](https://mlsanity.com/paper/2303.15343/) (2023)
- [Convex optimization](https://mlsanity.com/paper/2106.01946/) (2021)
- [Describing Textures in the Wild](https://mlsanity.com/paper/1311.3618/) (2013)
- [Equality of Opportunity in Supervised Learning](https://mlsanity.com/paper/1610.02413/) (2016)
- [SentencePiece: A simple and language independent subword tokenizer and detokenizer for Neural Text Processing](https://mlsanity.com/paper/1808.06226/) (2018)
- [A Survey on Bias and Fairness in Machine Learning](https://mlsanity.com/paper/1908.09635/) (2019)
- [LAION-400M: Open Dataset of CLIP-Filtered 400 Million Image-Text Pairs](https://mlsanity.com/paper/2111.02114/) (2021)
- [Microsoft COCO Captions: Data Collection and Evaluation Server](https://mlsanity.com/paper/1504.00325/) (2015)
- [Man is to Computer Programmer as Woman is to Homemaker? Debiasing Word Embeddings](https://mlsanity.com/paper/1607.06520/) (2016)
- [Scaling Autoregressive Models for Content-Rich Text-to-Image Generation](https://mlsanity.com/paper/2206.10789/) (2022)
- [CoCa: Contrastive Captioners are Image-Text Foundation Models](https://mlsanity.com/paper/2205.01917/) (2022)
- [Scaling Vision Transformers](https://mlsanity.com/paper/2106.04560/) (2021)
- [Semantics derived automatically from language corpora contain human-like biases](https://mlsanity.com/paper/1608.07187/) (2016)
- [Florence: A New Foundation Model for Computer Vision](https://mlsanity.com/paper/2111.11432/) (2021)
- [PaLI: A Jointly-Scaled Multilingual Language-Image Model](https://mlsanity.com/paper/2209.06794/) (2022)
- [Certifying and removing disparate impact](https://mlsanity.com/paper/1412.3756/) (2014)
- [A Reductions Approach to Fair Classification](https://mlsanity.com/paper/1803.02453/) (2018)
- [Inherent Trade-Offs in the Fair Determination of Risk Scores](https://mlsanity.com/paper/1609.05807/) (2016)
- [LiT: Zero-Shot Transfer with Locked-image text Tuning](https://mlsanity.com/paper/2111.07991/) (2021)
- [Age Progression/Regression by Conditional Adversarial Autoencoder](https://mlsanity.com/paper/1702.08423/) (2017)
- [Men Also Like Shopping: Reducing Gender Bias Amplification using Corpus-level Constraints](https://mlsanity.com/paper/1707.09457/) (2017)
- [Last Layer Re-Training is Sufficient for Robustness to Spurious Correlations](https://mlsanity.com/paper/2204.02937/) (2022)
- [Fairness Beyond Disparate Treatment & Disparate Impact: Learning Classification without Disparate Mistreatment](https://mlsanity.com/paper/1610.08452/) (2016)

## Built on by (most cited first, 30 of 35)

- [MLLMs are Deeply Affected by Modality Bias](https://mlsanity.com/paper/2505.18657/) (2025)
- [BIOMEDICA: An Open Biomedical Image-Caption Archive, Dataset, and Vision-Language Models Derived from Scientific Literature](https://mlsanity.com/paper/2501.07171/) (2025)
- [A CLIP-Powered Framework for Robust and Generalizable Data Selection](https://mlsanity.com/paper/2410.11215/) (2024)
- [SANER: Annotation-free Societal Attribute Neutralizer for Debiasing CLIP](https://mlsanity.com/paper/2408.10202/) (2024)
- [Scaling Pre-training to One Hundred Billion Data for Vision Language Models](https://mlsanity.com/paper/2502.07617/) (2025)
- [BendVLM: Test-Time Debiasing of Vision-Language Embeddings](https://mlsanity.com/paper/2411.04420/) (2024)
- [No Filter: Cultural and Socioeconomic Diversity in Contrastive Vision-Language Models](https://mlsanity.com/paper/2405.13777/) (2024)
- [Fairness and Bias Mitigation in Computer Vision: A Survey](https://mlsanity.com/paper/2408.02464/) (2024)
- [AstroM3: A self-supervised multimodal model for astronomy](https://mlsanity.com/paper/2411.08842/) (2024)
- [Mind the Graph When Balancing Data for Fairness or Robustness](https://mlsanity.com/paper/2406.17433/) (2024)
- [Multimodal Unlearning Across Vision, Language, Video, and Audio: Survey of Methods, Datasets, and Benchmarks](https://mlsanity.com/paper/2607.07907/) (2026)
- [Enhancing Concept Localization in CLIP-based Concept Bottleneck Models](https://mlsanity.com/paper/2510.07115/) (2025)
- [G2D: Boosting Multimodal Learning with Gradient-Guided Distillation](https://mlsanity.com/paper/2506.21514/) (2025)
- [PRISM: Reducing Spurious Implicit Biases in Vision-Language Models with LLM-Guided Embedding Projection](https://mlsanity.com/paper/2507.08979/) (2025)
- [Measuring Agreeableness Bias in Multimodal Models](https://mlsanity.com/paper/2408.09111/) (2024)
- [Debiasing CLIP: Interpreting and Correcting Bias in Attention Heads](https://mlsanity.com/paper/2505.17425/) (2025)
- [Person-Centric Annotations of LAION-400M: Auditing Bias and Its Transfer to Models](https://mlsanity.com/paper/2510.03721/) (2025)
- [They're All Doctors: Synthesizing Diverse Counterfactuals to Mitigate Associative Bias](https://mlsanity.com/paper/2406.11331/) (2024)
- [Fair Foundation Models for Medical Image Analysis: Challenges and Perspectives](https://mlsanity.com/paper/2502.16841/) (2025)
- [LoRAverse: A Submodular Framework to Retrieve Diverse Adapters for Diffusion Models](https://mlsanity.com/paper/2510.15022/) (2025)
- [Semantic Approach to Quantifying the Consistency of Diffusion Model Image Generation](https://mlsanity.com/paper/2404.08799/) (2024)
- [Breaking Language Barriers or Reinforcing Bias? A Study of Gender and Racial Disparities in Multilingual Contrastive Vision Language Models](https://mlsanity.com/paper/2505.14160/) (2025)
- [Unsupervised Data Validation Methods for Efficient Model Training](https://mlsanity.com/paper/2410.07880/) (2024)
- [More is Less? A Simulation-Based Approach to Dynamic Interactions between Biases in Multimodal Models](https://mlsanity.com/paper/2412.17505/) (2024)
- [SEM: Sparse Embedding Modulation for Post-Hoc Debiasing of Vision-Language Models](https://mlsanity.com/paper/2603.19028/) (2026)
- [A Critical Review of Predominant Bias in Neural Networks](https://mlsanity.com/paper/2502.11031/) (2025)
- [EMoE: Training-Free Expert Disagreement for Uncertainty-Aware Text-to-Image Diffusion](https://mlsanity.com/paper/2505.13273/) (2025)
- [Bias Analysis in Unconditional Image Generative Models](https://mlsanity.com/paper/2506.09106/) (2025)
- [Representation Invariance and Allocation: When Subgroup Balance Matters](https://mlsanity.com/paper/2512.09496/) (2025)
- [Hierarchical Pre-Training of Vision Encoders with Large Language Model](https://mlsanity.com/paper/2604.00086/) (2026)

## Abstract (arXiv)

We study the effectiveness of data-balancing for mitigating biases in contrastive language-image pretraining (CLIP), identifying areas of strength and limitation. First, we reaffirm prior conclusions that CLIP models can inadvertently absorb societal stereotypes. To counter this, we present a novel algorithm, called Multi-Modal Moment Matching (M4), designed to reduce both representation and association biases (i.e. in first- and second-order statistics) in multimodal data. We use M4 to conduct an in-depth analysis taking into account various factors, such as the model, representation, and data size. Our study also explores the dynamic nature of how CLIP learns and unlearns biases. In particular, we find that fine-tuning is effective in countering representation biases, though its impact diminishes for association biases. Also, data balancing has a mixed impact on quality: it tends to improve classification but can hurt retrieval. Interestingly, data and architectural improvements seem to mitigate the negative impact of data balancing on performance; e.g. applying M4 to SigLIP-B/16 with data quality filters improves COCO image-to-text retrieval @5 from 86% (without data balancing) to 87% and ImageNet 0-shot classification from 77% to 77.5%! Finally, we conclude with recommendations for improving the efficacy of data balancing in multimodal systems.