Summary of Da-net: a Disentangled and Adaptive Network For Multi-source Cross-lingual Transfer Learning, by Ling Ge et al.
DA-Net: A Disentangled and Adaptive Network for Multi-Source Cross-Lingual Transfer Learning
by Ling Ge, Chunming Hu, Guanghui Ma, Jihong Liu, Hong Zhang
First submitted to arxiv on: 7 Mar 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary A novel multi-source cross-lingual transfer learning framework is proposed to address the challenges of task knowledge transfer from multiple labelled source languages to an unlabeled target language under language shift. The framework, called Disentangled and Adaptive Network (DA-Net), aims to mitigate mutual interference from multiple sources and alleviate the language gap between source-target language pairs. DA-Net consists of a feedback-guided collaborative disentanglement method that purifies input representations of classifiers and a class-aware parallel adaptation method that aligns class-level distributions for each source-target language pair. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper proposes a new approach to multi-source cross-lingual transfer learning, which is useful for transferring task knowledge from multiple labelled source languages to an unlabeled target language. This can help machines learn new tasks in a new language without being trained on that language specifically. The approach, called Disentangled and Adaptive Network (DA-Net), helps the model by purifying input representations of classifiers and aligning class-level distributions for each source-target language pair. |
Keywords
» Artificial intelligence » Transfer learning
