Multi-Prompt Denoised Self-Training for Open-Vocabulary Model Adaptation
Keywords: vision language models, model adaptation, transductive transfer learning
Abstract: Traditional model adaptation assumes the same vocabulary across source and target domains, which often struggles with limited transfer flexibility and efficiency while handling target domains with different vocabularies. Inspired by recent vision-language models (VLMs) that enable open-vocabulary visual recognition by reasoning on both images and texts, we study open-vocabulary model adaptation (OVMA), a new unsupervised model adaptation framework that positions a pre-trained VLM as the source model and transfers it towards arbitrary unlabelled target domains. To this end, we design a Multi-prompt denOised Self-Training (MOST) technique that exploits the synergy between vision and language to mitigate the domain discrepancies in image and text distributions simultaneously. Specifically, MOST makes use of the complementary property of multiple prompts within and across vision and language modalities, which enables joint exploitation of vision and language information and effective learning of image-text correspondences in the unlabelled target domains. Additionally, MOST captures temporal information via multi-temporal prompt learning which helps memorize previously learnt target information. Extensive experiments show that MOST outperforms the state-of-the-art consistently across 11 image recognition tasks. Codes will be released
Primary Area: transfer learning, meta learning, and lifelong learning
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 3095
Loading