Visually Grounded Continual Language Learning with Selective Specialization

Kyra Ahrens, Lennart Bengtson, Jae Hee Lee, Stefan Wermter


Abstract
A desirable trait of an artificial agent acting in the visual world is to continually learn a sequence of language-informed tasks while striking a balance between sufficiently specializing in each task and building a generalized knowledge for transfer. Selective specialization, i.e., a careful selection of model components to specialize in each task, is a strategy to provide control over this trade-off. However, the design of selection strategies requires insights on the role of each model component in learning rather specialized or generalizable representations, which poses a gap in current research. Thus, our aim with this work is to provide an extensive analysis of selection strategies for visually grounded continual language learning. Due to the lack of suitable benchmarks for this purpose, we introduce two novel diagnostic datasets that provide enough control and flexibility for a thorough model analysis. We assess various heuristics for module specialization strategies as well as quantifiable measures for two different types of model architectures. Finally, we design conceptually simple approaches based on our analysis that outperform common continual learning baselines. Our results demonstrate the need for further efforts towards better aligning continual learning algorithms with the learning behaviors of individual model parts.
Anthology ID:
2023.findings-emnlp.469
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2023
Month:
December
Year:
2023
Address:
Singapore
Editors:
Houda Bouamor, Juan Pino, Kalika Bali
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
7037–7054
Language:
URL:
https://aclanthology.org/2023.findings-emnlp.469
DOI:
10.18653/v1/2023.findings-emnlp.469
Bibkey:
Cite (ACL):
Kyra Ahrens, Lennart Bengtson, Jae Hee Lee, and Stefan Wermter. 2023. Visually Grounded Continual Language Learning with Selective Specialization. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 7037–7054, Singapore. Association for Computational Linguistics.
Cite (Informal):
Visually Grounded Continual Language Learning with Selective Specialization (Ahrens et al., Findings 2023)
Copy Citation:
PDF:
https://aclanthology.org/2023.findings-emnlp.469.pdf