Error Diversity Matters: An Error-Resistant Ensemble Method for Unsupervised Dependency Parsing
Abstract
We address unsupervised dependency parsing by building an ensemble of diverse existing models through post hoc aggregation of their output dependency parse structures. We observe that these ensembles often suffer from low robustness against weak ensemble components due to error accumulation. To tackle this problem, we propose an efficient ensemble-selection approach that considers error diversity and avoids error accumulation. Results demonstrate that our approach outperforms each individual model as well as previous ensemble techniques. Additionally, our experiments show that the proposed ensemble-selection method significantly enhances the performance and robustness of our ensemble, surpassing previously proposed strategies, which have not accounted for error diversity.
Cite
Text
Shayegh et al. "Error Diversity Matters: An Error-Resistant Ensemble Method for Unsupervised Dependency Parsing." AAAI Conference on Artificial Intelligence, 2025. doi:10.1609/AAAI.V39I23.34697Markdown
[Shayegh et al. "Error Diversity Matters: An Error-Resistant Ensemble Method for Unsupervised Dependency Parsing." AAAI Conference on Artificial Intelligence, 2025.](https://mlanthology.org/aaai/2025/shayegh2025aaai-error/) doi:10.1609/AAAI.V39I23.34697BibTeX
@inproceedings{shayegh2025aaai-error,
title = {{Error Diversity Matters: An Error-Resistant Ensemble Method for Unsupervised Dependency Parsing}},
author = {Shayegh, Behzad and Lee, Hobie H.-B. and Zhu, Xiaodan and Cheung, Jackie Chi Kit and Mou, Lili},
booktitle = {AAAI Conference on Artificial Intelligence},
year = {2025},
pages = {25119-25127},
doi = {10.1609/AAAI.V39I23.34697},
url = {https://mlanthology.org/aaai/2025/shayegh2025aaai-error/}
}