Learning to Learn Morphological Inflection for Resource-Poor Languages
Abstract
We propose to cast the task of morphological inflection—mapping a lemma to an indicated inflected form—for resource-poor languages as a meta-learning problem. Treating each language as a separate task, we use data from high-resource source languages to learn a set of model parameters that can serve as a strong initialization point for fine-tuning on a resource-poor target language. Experiments with two model architectures on 29 target languages from 3 families show that our suggested approach outperforms all baselines. In particular, it obtains a 31.7% higher absolute accuracy than a previously proposed cross-lingual transfer model and outperforms the previous state of the art by 1.7% absolute accuracy on average over languages.
Cite
Text
Kann et al. "Learning to Learn Morphological Inflection for Resource-Poor Languages." AAAI Conference on Artificial Intelligence, 2020. doi:10.1609/AAAI.V34I05.6316Markdown
[Kann et al. "Learning to Learn Morphological Inflection for Resource-Poor Languages." AAAI Conference on Artificial Intelligence, 2020.](https://mlanthology.org/aaai/2020/kann2020aaai-learning/) doi:10.1609/AAAI.V34I05.6316BibTeX
@inproceedings{kann2020aaai-learning,
title = {{Learning to Learn Morphological Inflection for Resource-Poor Languages}},
author = {Kann, Katharina and Bowman, Samuel R. and Cho, Kyunghyun},
booktitle = {AAAI Conference on Artificial Intelligence},
year = {2020},
pages = {8058-8065},
doi = {10.1609/AAAI.V34I05.6316},
url = {https://mlanthology.org/aaai/2020/kann2020aaai-learning/}
}