面向组合泛化的组合增量学习
Composition-Incremental Learning for Compositional Generalization
AI总结：
本文针对现实数据持续涌现、组合近乎无限等问题，在组合零样本学习任务中探索组合增量学习，构建MIT-States-CompIL等基准，提出含视觉合成器和语言基元蒸馏的伪重放框架，提升模型组合泛化能力。
摘要：
组合泛化在计算机视觉的预收集训练数据上已取得显著进展。然而，现实世界的数据持续涌现，其可能的组合近乎无限、呈长尾分布且无法完全可见。因此，理想模型应能以增量方式逐步提升组合泛化能力。本文在组合零样本学习（CZSL）任务背景下探索面向组合泛化的组合增量学习（CompIL），模型需持续学习新组合，以逐步提升组合泛化能力。为定量评估CompIL，我们利用现有数据集开发基准构建流程，得到MIT-States-CompIL和C-GQA-CompIL。此外，我们提出伪重放框架，该框架使用视觉合成器合成已学组合的视觉表示，并通过语言基元蒸馏机制在学习过程中保持对齐的基元表示。大量实验证明了所提框架的有效性。
英文摘要：
Compositional generalization has achieved substantial progress in computer vision on pre-collected training data. Nonetheless, real-world data continually emerges, with possible compositions being nearly infinite, long-tailed, and not entirely visible. Thus, an ideal model is supposed to gradually improve the capability of compositional generalization in an incremental manner. In this paper, we explore Composition-Incremental Learning for Compositional Generalization (CompIL) in the context of the compositional zero-shot learning (CZSL) task, where models need to continually learn new compositions, intending to improve their compositional generalization capability progressively. To quantitatively evaluate CompIL, we develop a benchmark construction pipeline leveraging existing datasets, yielding MIT-States-CompIL and C-GQA-CompIL. Furthermore, we propose a pseudo-replay framework utilizing a visual synthesizer to synthesize visual representations of learned compositions and a linguistic primitive distillation mechanism to maintain aligned primitive representations across the learning process. Extensive experiments demonstrate the effectiveness of the proposed framework.
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。
