神经机器翻译模型用于低资源的哈萨克语-英语语言对
Vladislav Karyukin1, Diana Rakhimova1,2, Aidana Karibayeva1
1Department of Information Systems, Al-Farabi Kazakh National University, Almaty, Kazakhstan.
PeerJ. Computer science
|June 22, 2023
概括
由于数据有限,神经机器翻译与像哈萨克语这样的低资源语言扎. 这项研究使用前向翻译和先进的神经网络模型改进了哈萨克语-英语翻译,双向RNN和字节对编码取得了最佳结果.
科学领域:
- 计算语言学 计算语言学
- 自然语言处理自然语言处理.
- 人工智能的人工智能
背景情况:
- 机器翻译 (MT) 对全球沟通至关重要,神经MT (NMT) 是主要的方法.
- 像哈萨克语这样的低资源语言在NMT中面临挑战,因为平行语体的稀缺性.
- 现有的增强数据和提高低资源语言的NMT性能的方法包括向前翻译,向后翻译和转移学习.
研究的目的:
- 探索和实施用于改善低资源场景的哈萨克语-英语机器翻译性能的方法.
- 为哈萨克斯坦研究数据增强技术和不同NMT架构的有效性.
- 评估代币化策略对翻译质量的影响.
主要方法:
- 利用前向翻译从单语言哈萨克语数据创建合成并行句子.
- 结合了合成数据与政府网站的平行体,共计38万句.
- 训练的序列对序列模型,包括循环神经网络 (RNN),双向RNN (BRNN) 和变压器,使用OpenNMT框架.
- 采用字节对编码 (BPE) 和词代码化用于比较分析.
主要成果:
- 双向RNN (BRNN) 和RNN模型在翻译准确度方面表现优于变压器模型.
- 字节对编码 (BPE) 代码化与单词代码化相比,产生了更高的性能指标.
- BRNN和BPE的组合取得了最好的结果,BLEU得分为0.49,WER为0.51,TER为0.45.
结论:
- 数据增强技术,如前向翻译,对于低资源的NMT是有效的.
- 双向RNN架构与子词标记化 (BPE) 结合,为增强哈萨克语-英语翻译提供了一个有希望的方法.
- 该研究表明,对于具有有限并行数据的语言,改善MT质量的可行策略是可行的.
关键词:
BRNNN BRNN 在线观看逆向翻译是一种逆向翻译.英语英语英语英语英语英语英语英语转发翻译 转发翻译哈萨克斯坦的哈萨克斯坦人哈萨克斯坦人哈萨克斯坦人神经机器翻译 神经机器翻译开放NMT是什么意思一个RNN RNNSeq2SeqSeq Seq2SeqSeq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq Seq2Seq2Seq3 Seq2Seq2Seq3 Seq2Seq4 Seq2Seq3 Seq2Seq4 Seq2Seq3 Seq2Seq4 Seq3 Seq2Seq变压器变压器变压器
相关概念视频
02:07Improving Translational Accuracy
2.6K
2.6K
01:31Translation
15.0K
Translation is the process of synthesizing proteins from the genetic information carried by messenger RNA (mRNA). Following transcription, it constitutes the final step in the expression of genes. This process is carried out by ribosomes, complexes of protein and specialized RNA molecules. Ribosomes, transfer RNA (tRNA), and other proteins produce a chain of amino acids—the polypeptide—as the end product of translation.
Translation Produces the Building Blocks of Life
Proteins are...
Translation Produces the Building Blocks of Life
Proteins are...
15.0K
02:28Leaky Scanning
5.2K
During most eukaryotic translation processes, the small 40S ribosome subunit scans an mRNA from its 5' end until it encounters the first start AUG codon. The large 60S ribosomal subunit then joins the smaller one to initiate protein synthesis. The location of the translation initiation is largely determined by the nucleotides near the start codon as there may be multiple translation initiation sites present on the mRNA. Marilyn Kozak discovered that the sequence RCCAUGG (where R...
5.2K
01:30Simplified Synchronous Machine Model
288
The Synchronous Machine Model is a fundamental tool in analyzing and ensuring the transient stability of power systems. This model simplifies the representation of a synchronous machine under balanced three-phase positive-sequence conditions, assuming constant excitation and ignoring losses and saturation. The model is pivotal for understanding the behavior of synchronous generators connected to a power grid, particularly during transient events.
In this model, each generator is connected to a...
In this model, each generator is connected to a...
288
02:33Initiation of Translation
6.5K
6.5K
01:14Mechanical Efficiency of Real Machines
763
The mechanical efficiency of a machine is a fundamental concept that describes how effectively a machine can convert input work into output work. According to this concept, the efficiency of a machine is equal to the ratio of the output work to the input work. An ideal machine, meaning a machine that has no energy losses, has an efficiency of one. This implies that the input work and the output work are equal.
However, in reality, no machine can be truly ideal, and all of them experience some...
However, in reality, no machine can be truly ideal, and all of them experience some...
763