CTP：通过兼容动量对比与拓扑保持迈向视觉-语言持续预训练
CTP: Towards Vision-Language Continual Pretraining via Compatible Momentum Contrast and Topology Preservation
AI总结：
针对视觉-语言预训练难以持续积累知识的问题，本文构建含9个行业、逾百万图文对的P9D基准，并提出CTP算法，通过兼容动量对比与拓扑保持实现高效VLCP。
摘要：
视觉-语言预训练（VLP）通过在大规模数据集上进行离线训练，已在多种下游任务上取得令人瞩目的结果。考虑到现实世界数据不断增长的特性，这种在持续扩张的数据上采用的离线训练范式是不可持续的，因为模型缺乏不断积累知识的持续学习能力。然而，大多数持续学习研究局限于单模态分类，并且现有多模态数据集无法模拟持续的非平稳数据流场景。为支持视觉-语言持续预训练（VLCP）研究，我们首先贡献了一个全面且统一的基准数据集 P9D，其中包含来自 9 个行业的超过一百万条商品图文对。每个行业的数据作为一项独立任务，支持持续学习，并符合现实世界的长尾特性，以模拟在网络数据上进行预训练。我们全面研究了 VLCP 的特征与挑战，并提出一种新算法：兼容动量对比与拓扑保持，命名为 CTP。兼容动量模型吸收当前任务模型和先前任务模型的知识，以灵活更新模态特征。此外，拓扑保持在跨任务迁移嵌入知识的同时，保持特征调整的灵活性。实验结果表明，与其他基线相比，我们的方法不仅取得了更优性能，而且不会带来昂贵的训练负担。数据集和代码可在 https://github.com/KevinLight831/CTP 获取。
英文摘要：
Vision-Language Pretraining (VLP) has shown impressive results on diverse downstream tasks by offline training on large-scale datasets. Regarding the growing nature of real-world data, such an offline training paradigm on ever-expanding data is unsustainable, because models lack the continual learning ability to accumulate knowledge constantly. However, most continual learning studies are limited to uni-modal classification and existing multi-modal datasets cannot simulate continual non-stationary data stream scenarios. To support the study of Vision-Language Continual Pretraining (VLCP), we first contribute a comprehensive and unified benchmark dataset P9D which contains over one million product image-text pairs from 9 industries. The data from each industry as an independent task supports continual learning and conforms to the real-world long-tail nature to simulate pretraining on web data. We comprehensively study the characteristics and challenges of VLCP, and propose a new algorithm: Compatible momentum contrast with Topology Preservation, dubbed CTP. The compatible momentum model absorbs the knowledge of the current and previous-task models to flexibly update the modal feature. Moreover, Topology Preservation transfers the knowledge of embedding across tasks while preserving the flexibility of feature adjustment. The experimental results demonstrate our method not only achieves superior performance compared with other baselines but also does not bring an expensive training burden. Dataset and codes are available at https://github.com/KevinLight831/CTP.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
