FewShotTextGCN: K-hop neighborhood regularization for few-shot learning on graphs
This addresses low-resource document classification for typologically diverse languages without requiring pretraining, though it is incremental in graph-based methods.
The paper tackled few-shot learning on word-document graphs by introducing K-hop neighborhood regularization and a simplified graph construction, achieving a 17% absolute accuracy improvement over a baseline with only 20 training samples across eight languages.
We present FewShotTextGCN, a novel method designed to effectively utilize the properties of word-document graphs for improved learning in low-resource settings. We introduce K-hop Neighbourhood Regularization, a regularizer for heterogeneous graphs, and show that it stabilizes and improves learning when only a few training samples are available. We furthermore propose a simplification in the graph-construction method, which results in a graph that is $\sim$7 times less dense and yields better performance in little-resource settings while performing on par with the state of the art in high-resource settings. Finally, we introduce a new variant of Adaptive Pseudo-Labeling tailored for word-document graphs. When using as little as 20 samples for training, we outperform a strong TextGCN baseline with 17% in absolute accuracy on average over eight languages. We demonstrate that our method can be applied to document classification without any language model pretraining on a wide range of typologically diverse languages while performing on par with large pretrained language models.