Uncovering Bias in Large Vision-Language Models at Scale with Counterfactuals

• URL: http://arxiv.org/abs/2405.20152v1
• Date: Thu, 30 May 2024 15:27:56 GMT
• Title: Uncovering Bias in Large Vision-Language Models at Scale with Counterfactuals
• Authors: Phillip Howard, Kathleen C. Fraser, Anahita Bhiwandiwalla, Svetlana Kiritchenko,
• Abstract summary: We study the social biases contained in text generated by Large Vision-Language Models (LVLMs) We present LVLMs with identical open-ended text prompts while conditioning on images from different counterfactual sets. We evaluate the text produced by different models under this counterfactual generation setting at scale, producing over 57 million responses from popular LVLMs.
• Score: 8.41410889524315
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: With the advent of Large Language Models (LLMs) possessing increasingly impressive capabilities, a number of Large Vision-Language Models (LVLMs) have been proposed to augment LLMs with visual inputs. Such models condition generated text on both an input image and a text prompt, enabling a variety of use cases such as visual question answering and multimodal chat. While prior studies have examined the social biases contained in text generated by LLMs, this topic has been relatively unexplored in LVLMs. Examining social biases in LVLMs is particularly challenging due to the confounding contributions of bias induced by information contained across the text and visual modalities. To address this challenging problem, we conduct a large-scale study of text generated by different LVLMs under counterfactual changes to input images. Specifically, we present LVLMs with identical open-ended text prompts while conditioning on images from different counterfactual sets, where each set contains images which are largely identical in their depiction of a common subject (e.g., a doctor), but vary only in terms of intersectional social attributes (e.g., race and gender). We comprehensively evaluate the text produced by different models under this counterfactual generation setting at scale, producing over 57 million responses from popular LVLMs. Our multi-dimensional analysis reveals that social attributes such as race, gender, and physical characteristics depicted in input images can significantly influence the generation of toxic content, competency-associated words, harmful stereotypes, and numerical ratings of depicted individuals. We additionally explore the relationship between social bias in LVLMs and their corresponding LLMs, as well as inference-time strategies to mitigate bias.

Related papers

• Rethinking Visual Prompting for Multimodal Large Language Models with External Knowledge [76.45868419402265]
  multimodal large language models (MLLMs) have made significant strides by training on vast high-quality image-text datasets. However, the inherent difficulty in explicitly conveying fine-grained or spatially dense information in text, such as masks, poses a challenge for MLLMs. This paper proposes a new visual prompt approach to integrate fine-grained external knowledge, gleaned from specialized vision models, into MLLMs.
  arXiv  Detail & Related papers  (2024-07-05T17:43:30Z)
• Social Bias Evaluation for Large Language Models Requires Prompt Variations [38.91306092184724]
  Large Language Models (LLMs) exhibit considerable social biases. This paper investigates the sensitivity of LLMs when changing prompt variations. We show that LLMs have tradeoffs between performance and social bias caused by the prompts.
  arXiv  Detail & Related papers  (2024-07-03T14:12:04Z)
• Evaluating Fairness in Large Vision-Language Models Across Diverse Demographic Attributes and Prompts [27.66626125248612]
  We empirically investigate emphvisual fairness in several mainstream vision-language models (LVLMs) We audit their performance disparities across sensitive demographic attributes based on public fairness benchmark datasets (e.g., FACET) Despite enhancements in visual understanding, both open-source and closed-source LVLMs exhibit prevalent fairness issues across different instruct prompts and demographic attributes.
  arXiv  Detail & Related papers  (2024-06-25T23:11:39Z)
• Implicit Multimodal Alignment: On the Generalization of Frozen LLMs to Multimodal Inputs [63.29737699997859]
  Large Language Models (LLMs) have demonstrated impressive performance on multimodal tasks, without any multimodal finetuning. In this work, we expose frozen LLMs to image, video, audio and text inputs and analyse their internal representation.
  arXiv  Detail & Related papers  (2024-05-26T21:31:59Z)
• Uncovering Bias in Large Vision-Language Models with Counterfactuals [8.414108895243148]
  We study the social biases contained in text generated by Large Vision-Language Models (LVLMs) We present LVLMs with identical open-ended text prompts while conditioning on images from different counterfactual sets. We find that social attributes such as race, gender, and physical characteristics depicted in input images can significantly influence toxicity and the generation of competency-associated words.
  arXiv  Detail & Related papers  (2024-03-29T21:45:53Z)
• Debiasing Multimodal Large Language Models [61.6896704217147]
  Large Vision-Language Models (LVLMs) have become indispensable tools in computer vision and natural language processing. Our investigation reveals a noteworthy bias in the generated content, where the output is primarily influenced by the underlying Large Language Models (LLMs) prior to the input image. To rectify these biases and redirect the model's focus toward vision information, we introduce two simple, training-free strategies.
  arXiv  Detail & Related papers  (2024-03-08T12:35:07Z)
• A Unified Framework and Dataset for Assessing Societal Bias in Vision-Language Models [9.025958469582363]
  We propose a unified framework for evaluating gender, race, and age biases in vision-language models (VLMs) We generate high-quality synthetic datasets that intentionally conceal gender, race, and age information across different professional domains. The dataset includes action-based descriptions of each profession and serves as a benchmark for evaluating societal biases in vision-language models (VLMs)
  arXiv  Detail & Related papers  (2024-02-21T09:17:51Z)
• InternLM-XComposer2: Mastering Free-form Text-Image Composition and Comprehension in Vision-Language Large Model [108.42241250772643]
  We introduce InternLM-XComposer2, a vision-language model excelling in free-form text-image composition and comprehension. This model goes beyond conventional vision-language understanding, adeptly crafting interleaved text-image content from diverse inputs. Experimental results demonstrate the superiority of InternLM-XComposer2 based on InternLM2-7B in producing high-quality long-text multi-modal content.
  arXiv  Detail & Related papers  (2024-01-29T18:59:02Z)
• Mitigating Hallucination in Visual Language Models with Visual Supervision [33.05550629039951]
  Large vision-language models (LVLMs) suffer from hallucination a lot. Key problem lies in its weak ability to comprehend detailed content in a multi-modal context. In this paper, we bring more detailed vision annotations and more discriminative vision models to facilitate the training of LVLMs.
  arXiv  Detail & Related papers  (2023-11-27T09:30:02Z)
• TouchStone: Evaluating Vision-Language Models by Language Models [91.69776377214814]
  We propose an evaluation method that uses strong large language models as judges to comprehensively evaluate the various abilities of LVLMs. We construct a comprehensive visual dialogue dataset TouchStone, consisting of open-world images and questions, covering five major categories of abilities and 27 subtasks. We demonstrate that powerful LVLMs, such as GPT-4, can effectively score dialogue quality by leveraging their textual capabilities alone.
  arXiv  Detail & Related papers  (2023-08-31T17:52:04Z)
• Visually-Augmented Language Modeling [137.36789885105642]
  We propose a novel pre-training framework, named VaLM, to Visually-augment text tokens with retrieved relevant images for Language Modeling. With the visually-augmented context, VaLM uses a visual knowledge fusion layer to enable multimodal grounded language modeling. We evaluate the proposed model on various multimodal commonsense reasoning tasks, which require visual information to excel.
  arXiv  Detail & Related papers  (2022-05-20T13:41:12Z)

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.