Scholarlink

Several approaches to solving challenges with limited resources in neural machine translation are explored in this survey. In particular, the case of English-Kannada NMT is used as an example. For NMT systems to make good translations, they need a lot of parallel corpora. The English-Hindi parallel corpus is used by the one-to-many (English to Hindi and Malayalam) multilingual model to enhance the quality of English-Kannada translations. For improving parallel data, techniques like phrase table injection, back-translation, and combined corpus methods are used. For improving transfer learning, techniques like pivoting and multilingual embeddings are used. When translating from English to Kannada, Hindi can be used to help with pivoting. In this article, a detailed case study is done with respect to machine translation between English and a Kannada language. The various Machine translation evaluation metrics are reviewed.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Neural Machine Translation in Low-Resource Context: Survey

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Padma Prasada,
  • M. V. Panduranga Rao,
  • Ujwala Vishwanatharao Suryawanshi

摘要

Several approaches to solving challenges with limited resources in neural machine translation are explored in this survey. In particular, the case of English-Kannada NMT is used as an example. For NMT systems to make good translations, they need a lot of parallel corpora. The English-Hindi parallel corpus is used by the one-to-many (English to Hindi and Malayalam) multilingual model to enhance the quality of English-Kannada translations. For improving parallel data, techniques like phrase table injection, back-translation, and combined corpus methods are used. For improving transfer learning, techniques like pivoting and multilingual embeddings are used. When translating from English to Kannada, Hindi can be used to help with pivoting. In this article, a detailed case study is done with respect to machine translation between English and a Kannada language. The various Machine translation evaluation metrics are reviewed.