Composition-Incremental Learning for Compositional Generalization
Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (8) · 6735-6743
DOI: 10.1609/aaai.v40i8.37605 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Zhen Li und 5 weiteren Autor:innen.
Abstract (Originalsprache)
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.
Zitieren
APA
Li, Z., Wu, Y., Jing, C., Sun, C., Li, C., & Jia, Y. (2026). Composition-Incremental Learning for Compositional Generalization. Proceedings of the AAAI Conference on Artificial Intelligence, 40(8), 6735-6743. https://doi.org/10.1609/aaai.v40i8.37605
Vancouver
Li Z, Wu Y, Jing C, Sun C, Li C, Jia Y. Composition-Incremental Learning for Compositional Generalization. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(8):6735-6743. doi:10.1609/aaai.v40i8.37605
BibTeX
@article{Li2026,
author = {Li, Zhen and Wu, Yuwei and Jing, Chenchen and Sun, Che and Li, Chuanhao and Jia, Yunde},
title = {Composition-Incremental Learning for Compositional Generalization},
journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
year = {2026},
volume = {40},
number = {8},
pages = {6735--6743},
doi = {10.1609/aaai.v40i8.37605}
}
EUVIMED