Existing approaches for unsupervised bilingual lexicon induction (BLI) often
depend on good quality static or contextual embeddings trained on large
monolingual corpora for both languages. In reality, however, unsupervised BLI
is most likely to be useful for dialects and languages that do not have
abundant amounts of monolingual data. We introduce a simple and fast method for
unsupervised BLI for low-resource languages with a related mid-to-high resource
language, only requiring inference on the higher-resource language monolingual
BERT. We work with two low-resource languages ($<5M$ monolingual tokens),
Bhojpuri and Magahi, of the severely under-researched Indic dialect continuum,
showing that state-of-the-art methods in the literature show near-zero
performance in these settings, and that our simpler method gives much better
results. We repeat our experiments on Marathi and Nepali, two higher-resource
Indic languages, to compare approach performances by resource range. We release
automatically created bilingual lexicons for the first time for five languages
of the Indic dialect continuum.

介绍了一种适用于资源匮乏语言的无监督双语词汇表感知方法，该方法仅需要使用高资源语言单语 BERT 进行推理，并在印度次大陆的五种语言中首次发布自动生成的双语词汇表。

一种简单的方法用于数据不平衡、紧密相关的语言对的无监督双语词典归纳

A Simple Method for Unsupervised Bilingual Lexicon Induction for  Data-Imbalanced, Closely Related Language Pairs

Generative adversarial networks (GANs) have succeeded in inducing
cross-lingual word embeddings -- maps of matching words across languages --
without supervision. Despite these successes, GANs' performance for the
difficult case of distant languages is still not satisfactory. These
limitations have been explained by GANs' incorrect assumption that source and
target embedding spaces are related by a single linear mapping and are
approximately isomorphic. We assume instead that, especially across distant
languages, the mapping is only piece-wise linear, and propose a
multi-adversarial learning method. This novel method induces the seed
cross-lingual dictionary through multiple mappings, each induced to fit the
mapping for one subspace. Our experiments on unsupervised bilingual lexicon
induction show that this method improves performance over previous
single-mapping methods, especially for distant languages.

提出了一种基于多对抗学习的新方法，用于在远程语言之间诱导跨语言词嵌入，改善了以往单映射方法的性能，特别是对于远程语言的情况

多重对抗学习用于跨语言单词嵌入

Multi-Adversarial Learning for Cross-Lingual Word Embeddings

Recent embedding-based methods in unsupervised bilingual lexicon induction
have shown good results, but generally have not leveraged orthographic
(spelling) information, which can be helpful for pairs of related languages.
This work augments a state-of-the-art method with orthographic features, and
extends prior work in this space by proposing methods that can learn and
utilize orthographic correspondences even between languages with different
scripts. We demonstrate this by experimenting on three language pairs with
different scripts and varying degrees of lexical similarity.

本文研究了使用无监督双语词汇识别中基于嵌入的最新方法，提出了一种使用拼写信息的强化方法，并提出了一种可以学习和利用具有不同语言间的正字对应关系的方法。实验结果表明该方法在使用不同文字和不同词汇相似度的三种语言对上表现出了很好的效果。