Interpreting and Editing Vision-Language Representations to Mitigate Hallucinations

• URL: http://arxiv.org/abs/2410.02762v2
• Date: Mon, 10 Feb 2025 20:13:31 GMT
• Title: Interpreting and Editing Vision-Language Representations to Mitigate Hallucinations
• Authors: Nick Jiang, Anish Kachinthaya, Suzie Petryk, Yossi Gandelsman,
• Abstract summary: We investigate the internal representations of vision-language models (VLMs) to address hallucinations. We project VLMs' internal image representations to their language vocabulary and observe more confident output probabilities on real objects than hallucinated objects. We show that targeted edits to a model's latent representations can reduce hallucinations by up to 25.7% on the COCO2014 dataset.
• Score: 15.035663040732798
• License:
• Abstract: We investigate the internal representations of vision-language models (VLMs) to address hallucinations, a persistent challenge despite advances in model size and training. We project VLMs' internal image representations to their language vocabulary and observe more confident output probabilities on real objects than hallucinated objects. We additionally use these output probabilities to spatially localize real objects. Building on this approach, we introduce a knowledge erasure algorithm that removes hallucinations by linearly orthogonalizing image features with respect to hallucinated object features. We show that targeted edits to a model's latent representations can reduce hallucinations by up to 25.7% on the COCO2014 dataset while preserving performance. Our findings demonstrate how a deeper understanding of VLMs' latent representations can enhance reliability and enable novel capabilities, such as zero-shot segmentation.

Related papers

• Self-Correcting Decoding with Generative Feedback for Mitigating Hallucinations in Large Vision-Language Models [66.71616369573715]
  Large Vision-Language Models (LVLMs) are prone to generating hallucinatory text responses that do not align with the given visual input. We introduce self-correcting Decoding with Generative Feedback (DeGF), a novel training-free algorithm that incorporates feedback from text-to-image generative models into the decoding process.
  arXiv  Detail & Related papers  (2025-02-10T03:43:55Z)
• Evaluating Hallucination in Large Vision-Language Models based on Context-Aware Object Similarities [5.602853217226167]
  We present Context-Aware Object Similarities (CAOS), a novel approach for evaluating object hallucination in large vision-language models (LVLMs) CAOS integrates object statistics with semantic relationships between objects in captions and ground-truth data. To address this, we further employ language model-based object recognition to detect potentially out-of-domain hallucinated objects.
  arXiv  Detail & Related papers  (2025-01-25T03:03:18Z)
• HALLUCINOGEN: A Benchmark for Evaluating Object Hallucination in Large Visual-Language Models [57.58426038241812]
  Large Vision-Language Models (LVLMs) have demonstrated remarkable performance in performing complex multimodal tasks. We propose HALLUCINOGEN, a novel visual question answering (VQA) object hallucination attack benchmark. We extend our benchmark to high-stakes medical applications and introduce MED-HALLUCINOGEN, hallucination attacks tailored to the biomedical domain.
  arXiv  Detail & Related papers  (2024-12-29T23:56:01Z)
• Cracking the Code of Hallucination in LVLMs with Vision-aware Head Divergence [69.86946427928511]
  We investigate the internal mechanisms driving hallucination in large vision-language models (LVLMs) We introduce Vision-aware Head Divergence (VHD), a metric that quantifies the sensitivity of attention head outputs to visual context. We propose Vision-aware Head Reinforcement (VHR), a training-free approach to mitigate hallucination by enhancing the role of vision-aware attention heads.
  arXiv  Detail & Related papers  (2024-12-18T15:29:30Z)
• From Pixels to Tokens: Revisiting Object Hallucinations in Large Vision-Language Models [15.401221354325672]
  Hallucinations in large vision models (LVLMs) are a significant challenge, i.e., generating objects that are not presented in the visual input. Recent studies often attribute hallucinations to a lack of understanding of visual input, yet ignore a more fundamental issue: the model's inability to extract or decouple visual features. In this paper, we revisit the hallucinations in LVLMs from an architectural perspective, investigating whether the primary cause lies in the visual encoder (feature extraction) or the modal alignment module (feature decoupling)
  arXiv  Detail & Related papers  (2024-10-09T11:46:32Z)
• Investigating and Mitigating Object Hallucinations in Pretrained Vision-Language (CLIP) Models [22.42712853647949]
  We present an in-depth investigation into the object hallucination problem specifically within the CLIP model. We unveil that even in isolation, the CLIP model is prone to object hallucinations, suggesting that the hallucination problem is not solely due to the interaction between vision and language modalities. We show the the enhanced model can be employed as a visual encoder, effectively alleviating the object hallucination issue in LVLMs.
  arXiv  Detail & Related papers  (2024-10-04T06:24:49Z)
• Make VLM Recognize Visual Hallucination on Cartoon Character Image with Pose Information [5.501684533538496]
  We propose a novel semantic hallucination detection system using Vision-Language Model (VLM) We introduce pose-aware in-context visual learning (PA-ICVL) which improve the overall performance of VLM by further inputting visual data beyond prompts, RGB images and pose information. This research advances a capability of TTI models toward real-world applications by mitigating visual hallucinations via in-context visual learning, expanding their potential in non-photorealistic domains.
  arXiv  Detail & Related papers  (2024-03-22T09:13:09Z)
• Analyzing and Mitigating Object Hallucination in Large Vision-Language Models [110.12460299261531]
  Large vision-language models (LVLMs) have shown remarkable abilities in understanding visual information with human languages. LVLMs still suffer from object hallucination, which is the problem of generating descriptions that include objects that do not actually exist in the images. We propose a powerful algorithm, LVLM Hallucination Revisor (LURE), to rectify object hallucination in LVLMs by reconstructing less hallucinatory descriptions.
  arXiv  Detail & Related papers  (2023-10-01T18:10:53Z)
• Evaluating Object Hallucination in Large Vision-Language Models [122.40337582958453]
  This work presents the first systematic study on object hallucination of large vision-language models (LVLMs) We find that LVLMs tend to generate objects that are inconsistent with the target images in the descriptions. We propose a polling-based query method called POPE to evaluate the object hallucination.
  arXiv  Detail & Related papers  (2023-05-17T16:34:01Z)
• Plausible May Not Be Faithful: Probing Object Hallucination in Vision-Language Pre-training [66.0036211069513]
  Large-scale vision-language pre-trained models are prone to hallucinate non-existent visual objects when generating text. We show that models achieving better scores on standard metrics could hallucinate objects more frequently. Surprisingly, we find that patch-based features perform the best and smaller patch resolution yields a non-trivial reduction in object hallucination.
  arXiv  Detail & Related papers  (2022-10-14T10:27:22Z)

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.