Overcoming the Pitfalls of Vision-Language Model Finetuning for OOD Generalization

• URL: http://arxiv.org/abs/2401.15914v2
• Date: Tue, 16 Apr 2024 03:25:25 GMT
• Title: Overcoming the Pitfalls of Vision-Language Model Finetuning for OOD Generalization
• Authors: Yuhang Zang, Hanlin Goh, Josh Susskind, Chen Huang,
• Abstract summary: Existing vision-language models exhibit strong generalization on a variety of visual domains and tasks. We propose a novel approach OGEN to improve the OOD GENeralization of finetuned models. Specifically, a class-conditional feature generator is introduced to synthesize OOD features using just the class name of any unknown class.
• Score: 11.140366256534474
• License: http://creativecommons.org/licenses/by-sa/4.0/
• Abstract: Existing vision-language models exhibit strong generalization on a variety of visual domains and tasks. However, such models mainly perform zero-shot recognition in a closed-set manner, and thus struggle to handle open-domain visual concepts by design. There are recent finetuning methods, such as prompt learning, that not only study the discrimination between in-distribution (ID) and out-of-distribution (OOD) samples, but also show some improvements in both ID and OOD accuracies. In this paper, we first demonstrate that vision-language models, after long enough finetuning but without proper regularization, tend to overfit the known classes in the given dataset, with degraded performance on unknown classes. Then we propose a novel approach OGEN to address this pitfall, with the main focus on improving the OOD GENeralization of finetuned models. Specifically, a class-conditional feature generator is introduced to synthesize OOD features using just the class name of any unknown class. Such synthesized features will provide useful knowledge about unknowns and help regularize the decision boundary between ID and OOD data when optimized jointly. Equally important is our adaptive self-distillation mechanism to regularize our feature generation model during joint optimization, i.e., adaptively transferring knowledge between model states to further prevent overfitting. Experiments validate that our method yields convincing gains in OOD generalization performance in different settings. Code: https://github.com/apple/ml-ogen.

Related papers

• Effort: Efficient Orthogonal Modeling for Generalizable AI-Generated Image Detection [66.16595174895802]
  Existing AI-generated image (AIGI) detection methods often suffer from limited generalization performance. In this paper, we identify a crucial yet previously overlooked asymmetry phenomenon in AIGI detection.
  arXiv  Detail & Related papers  (2024-11-23T19:10:32Z)
• FlowCon: Out-of-Distribution Detection using Flow-Based Contrastive Learning [0.0]
  We introduce textitFlowCon, a new density-based OOD detection technique. Our main innovation lies in efficiently combining the properties of normalizing flow with supervised contrastive learning. Empirical evaluation shows the enhanced performance of our method across common vision datasets.
  arXiv  Detail & Related papers  (2024-07-03T20:33:56Z)
• CRoFT: Robust Fine-Tuning with Concurrent Optimization for OOD Generalization and Open-Set OOD Detection [42.33618249731874]
  We show that minimizing the magnitude of energy scores on training data leads to domain-consistent Hessians of classification loss. We have developed a unified fine-tuning framework that allows for concurrent optimization of both tasks.
  arXiv  Detail & Related papers  (2024-05-26T03:28:59Z)
• Secrets of RLHF in Large Language Models Part II: Reward Modeling [134.97964938009588]
  We introduce a series of novel methods to mitigate the influence of incorrect and ambiguous preferences in the dataset. We also introduce contrastive learning to enhance the ability of reward models to distinguish between chosen and rejected responses.
  arXiv  Detail & Related papers  (2024-01-11T17:56:59Z)
• Towards Calibrated Robust Fine-Tuning of Vision-Language Models [97.19901765814431]
  This work proposes a robust fine-tuning method that improves both OOD accuracy and confidence calibration simultaneously in vision language models. We show that both OOD classification and OOD calibration errors have a shared upper bound consisting of two terms of ID data. Based on this insight, we design a novel framework that conducts fine-tuning with a constrained multimodal contrastive loss enforcing a larger smallest singular value.
  arXiv  Detail & Related papers  (2023-11-03T05:41:25Z)
• Retrieval-Enhanced Contrastive Vision-Text Models [61.783728119255365]
  We propose to equip vision-text models with the ability to refine their embedding with cross-modal retrieved information from a memory at inference time. Remarkably, we show that this can be done with a light-weight, single-layer, fusion transformer on top of a frozen CLIP. Our experiments validate that our retrieval-enhanced contrastive (RECO) training improves CLIP performance substantially on several challenging fine-grained tasks.
  arXiv  Detail & Related papers  (2023-06-12T15:52:02Z)
• How Does Fine-Tuning Impact Out-of-Distribution Detection for Vision-Language Models? [29.75562085178755]
  We study how fine-tuning impact OOD detection for few-shot downstream tasks. Our results suggest that a proper choice of OOD scores is essential for CLIP-based fine-tuning. We also show that prompt learning demonstrates the state-of-the-art OOD detection performance over the zero-shot counterpart.
  arXiv  Detail & Related papers  (2023-06-09T17:16:50Z)
• Unleashing Mask: Explore the Intrinsic Out-of-Distribution Detection Capability [70.72426887518517]
  Out-of-distribution (OOD) detection is an indispensable aspect of secure AI when deploying machine learning models in real-world applications. We propose a novel method, Unleashing Mask, which aims to restore the OOD discriminative capabilities of the well-trained model with ID data. Our method utilizes a mask to figure out the memorized atypical samples, and then finetune the model or prune it with the introduced mask to forget them.
  arXiv  Detail & Related papers  (2023-06-06T14:23:34Z)
• Who Explains the Explanation? Quantitatively Assessing Feature Attribution Methods [0.0]
  We propose a novel evaluation metric -- the Focus -- designed to quantify the faithfulness of explanations. We show the robustness of the metric through randomization experiments, and then use Focus to evaluate and compare three popular explainability techniques. Our results find LRP and GradCAM to be consistent and reliable, while the latter remains most competitive even when applied to poorly performing models.
  arXiv  Detail & Related papers  (2021-09-28T07:10:24Z)
• OODformer: Out-Of-Distribution Detection Transformer [15.17006322500865]
  In real-world safety-critical applications, it is important to be aware if a new data point is OOD. This paper proposes a first-of-its-kind OOD detection architecture named OODformer.
  arXiv  Detail & Related papers  (2021-07-19T15:46:38Z)
• Enhancing the Generalization for Intent Classification and Out-of-Domain Detection in SLU [70.44344060176952]
  Intent classification is a major task in spoken language understanding (SLU) Recent works have shown that using extra data and labels can improve the OOD detection performance. This paper proposes to train a model with only IND data while supporting both IND intent classification and OOD detection.
  arXiv  Detail & Related papers  (2021-06-28T08:27:38Z)

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.