Learning to Learn Morphological Inflection for Resource-Poor Languages
This work addresses the challenge of limited data for morphological tasks in low-resource languages, offering an incremental improvement over existing methods.
The paper tackles morphological inflection for resource-poor languages by framing it as a meta-learning problem, using data from high-resource languages to initialize models for fine-tuning, resulting in a 31.7% higher accuracy than a cross-lingual transfer baseline and a 1.7% average improvement over previous state-of-the-art methods.
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.