Generalised Unsupervised Domain Adaptation of Neural Machine Translation with Cross-Lingual Data Selection

Most dominant neural machine translation (NMT) models are restricted to make predictions only according to the local context of preceding words in a left-to-right manner. Although many previous studies try to incorporate global information into NMT models, there still exist limitations on how to effectively exploit bidirectional global context. In this paper, we propose a Confidence Based Bidirectional Global Context Aware (CBBGCA) training framework for NMT, where the NMT model is jointly trained with an auxiliary conditional masked language model (CMLM). The training consists of two stages: (1) multi-task joint training; (2) confidence based knowledge distillation. At the first stage, by sharing encoder parameters, the NMT model is additionally supervised by the signal from the CMLM decoder that contains bidirectional global contexts. Moreover, at the second stage, using the CMLM as teacher, we further pertinently incorporate bidirectional global context to the NMT model on its unconfidently-predicted target words via knowledge distillation. Experimental results show that our proposed CBBGCA training framework significantly improves the NMT model by +1.02, +1.30 and +0.57 BLEU scores on three large-scale translation datasets, namely WMT'14 English-to-German, WMT'19 Chinese-to-English and WMT'14 English-to-French, respectively.

Machine translation aims to minimize the language barrier between people of different linguistic backgrounds. Machine translation is an automatic translation technique between pairs of different languages. Machine translation using neural network or neural machine translation came into the picture due to several limitation associated with its predecessors which are rule based and statistical based models. For large data sets with a rich range of vocabulary, the neural network machine translation system provides fair translation accuracy. We have observed that there remain very few machine translators dedicated to Indian languages, especially those spoken in the North-East area. So our paper mainly focuses on implementing a neural machine translation system for the English-Assamese language. In this paper we used five different neural machine translation models for English-Assamese language pair using LSTMs and GRUs, along with an attention layer. We have made a comparison analysis based on the performance results of these models. We used BLEU Score to calculate the accuracy of these five models, thus achieving a BLEU score of 34.168%, which was the highest among the five models.

The dominant neural machine translation (NMT) models that based on the encoder-decoder architecture have recently achieved the state-of-the-art performance. Traditionally, the NMT models only depend on the representations learned during training for mapping a source sentence into the target domain. However, the learned representations often suffer from implicit and inadequately informed properties. In this paper, we propose a novel bilingual topic enhanced NMT (BLTNMT) model to improve translation performance by incorporating bilingual topic knowledge into NMT. Specifically, the bilingual topic knowledge is included into the hidden states of both encoder and decoder, as well as the attention mechanism. With this new setting, the proposed BLT-NMT has access to the background knowledge implied in bilingual topics which is beyond the sequential context, and enables the attention mechanism to attend to topic-level attentions for generating accurate target words during translation. Experimental results show that the proposed model consistently outperforms the traditional RNNsearch and the previous topic-informed NMT on Chinese-English and EnglishGerman translation tasks. We also introduce the bilingual topic knowledge into the newly emerged Transformer base model on English-German translation and achieve a notable improvement.

Neural Machine Translation (NMT) has recently achieved the state-of-the-art in many machine translation tasks, but one of the challenges that NMT faces is the lack of parallel corpora, especially for low-resource language pairs. And the result is that the performance of NMT is much less effective for low-resource languages. To address this specific problem, in this paper, we describe a novel NMT model that is based on encoder-decoder architecture and relies on character-level inputs. Our proposed model employs Convolutional Neural Networks (CNN) and highway networks over character inputs, whose outputs are given to an encoder-decoder neural machine translation network. Besides, we also present two other approaches to improve the performance of the low-resource NMT system much further. First, we use language modeling implemented by denoising autoencoding to pre-train and initialize the full model. Second, we share the weights of the front few layers of two encoders between two languages to strengthen the encoding ability of the model. We demonstrate our model on two low-resource language pairs. On the IWSLT2015 English-Vietnamese translation task, our proposed model obtains improvements up to 2.5 BLEU points compared to the baseline. We also outperform the baseline approach more than 3 BLEU points on the CWMT2018 Chinese-Mongolian translation task.

Recent works have shown that Neural Machine Translation (NMT) models achieve impressive performance, however, questions about understanding the behavior of these models remain unanswered. We investigate the unexpected volatility of NMT models where the input is semantically and syntactically correct. We discover that with trivial modifications of source sentences, we can identify cases where unexpected changes happen in the translation and in the worst case lead to mistranslations. This volatile behavior of translating extremely similar sentences in surprisingly different ways highlights the underlying generalization problem of current NMT models. We find that both RNN and Transformer models display volatile behavior in 26% and 19% of sentence variations, respectively.

Neural machine translation (NMT) is sensitive to domain shift. In this paper, we address this problem in an active learning setting where we can spend a given budget on translating in-domain data, and gradually fine-tune a pre-trained out-of-domain NMT model on the newly translated data. Existing active learning methods for NMT usually select sentences based on uncertainty scores, but these methods require costly translation of full sentences even when only one or two key phrases within the sentence are informative. To address this limitation, we re-examine previous work from the phrase-based machine translation (PBMT) era that selected not full sentences, but rather individual phrases. However, while incorporating these phrases into PBMT systems was relatively simple, it is less trivial for NMT systems, which need to be trained on full sequences to capture larger structural properties of sentences unique to the new domain. To overcome these hurdles, we propose to select both full sentences and individual phrases from unlabelled data in the new domain for routing to human translators. In a German-English translation task, our active learning approach achieves consistent improvements over uncertainty-based sentence selection methods, improving up to 1.2 BLEU score over strong active learning baselines.

The attention mechanism performs well for the Neural Machine Translation (NMT) task, but heavily depends on the context vectors generated by the attention network to predict target words. This reliance raises the issue of long-term dependencies. Indeed, it is very common to combine predicates with postpositions in sentences, and the same predicate may have different meanings when combined with different postpositions. This usually poses an additional challenge to the NMT study. In this work, we observe that the embedding vectors of different target tokens can be classified by part-of-speech, thus we analyze the Natural Language Processing (NLP) related Content-Adaptive Recurrent Unit (CARU) unit and apply it to our attention model (CAAtt) and embedding layer (CAEmbed). By encoding the source sentence with the current decoded feature through the CARU, CAAtt is capable of achieving translation content-adaptive representations, which attention weights are contributed and enhanced by our proposed L1expNx normalization. Furthermore, CAEmbed aims to alleviate long-term dependencies in the target language through partial recurrent design, performing the feature extraction in a local perspective. Experiments on the WMT14, WMT17, and Multi30k translation tasks show that the proposed model achieves improvements in BLEU scores and enhancement of convergence over the attention-based plain NMT model. We also investigate the attention weights generated by the proposed approaches, which indicate that refinement over the different combinations of adposition can lead to different interpretations. Specifically, this work provides local attention to some specific phrases translated in our experiment. The results demonstrate that our approach is effective in improving performance and achieving a more reasonable attention distribution compared to the state-of-the-art models.

Most neural machine translation (NMT) systems rely on parallel data, comprising text in the source language and its corresponding translation in the target language. While it’s acknowledged that context enhances NMT models, this work proposes a novel approach by incorporating external context, specifically explanations of source text meanings, akin to how human translators leverage context for comprehension. The suggested methodology innovatively addresses the challenge of incorporating lengthy contextual information into NMT systems. By employing state-of-the-art transformer-based models, external context is integrated, thereby enriching the translation process. A key aspect of the approach lies in the utilization of diverse text summarization techniques, strategically employed to efficiently distill extensive contextual details into the NMT framework. This novel solution not only overcomes the obstacle posed by lengthy context but also enhances the translation quality, marking a advancement in the field of NMT. Furthermore, the data-centric approach ensures robustness and effectiveness, yielding improvements in translation quality, as evidenced by a considerable boost in BLEU score points ranging from 0.46 to 1.87 over baseline models. Additionally, we make our dataset publicly available, facilitating further research in this domain.

While end-to-end neural machine translation (NMT) has made remarkable progress recently, it still suffers from the data scarcity problem for low-resource language pairs and domains. In this paper, we propose a method for zero-resource NMT by assuming that parallel sentences have close probabilities of generating a sentence in a third language. Based on the assumption, our method is able to train a source-to-target NMT model ("student") without parallel corpora available guided by an existing pivot-to-target NMT model ("teacher") on a source-pivot parallel corpus. Experimental results show that the proposed method significantly improves over a baseline pivot-based model by +3.0 BLEU points across various language pairs.

The hypotactic structural relation between clauses plays an important role in improving the discourse coherence of document-level translation. However, the standard neural machine translation (NMT) models do not explicitly model the hypotactic relationship between clauses, which usually leads to structurally incorrect translations of long and complex sentences. This problem is particularly noticeable on Chinese-to-English translation task of complex sentences due to the grammatical form distinction between English and Chinese. English is rich in grammatical form (e.g. verb morphological changes and subordinating conjunctions) while Chinese is poor in grammatical form. These linguistic phenomena make it a challenge for NMT to learn the hypotactic structure knowledge from Chinese as well as the structure alignment between Chinese and English. To address these issues, we propose to model the hypotactic structure for Chinese-to-English complex sentence translation by introducing hypotactic structure knowledge. Specifically, we annotate and build a hypotactic structure aligned parallel corpus that provides rich hypotactic structure knowledge for NMT. Moreover, we further propose a structure-infused neural framework to combine the hypotactic structure knowledge with the NMT model through two integrating strategies. In particular, we introduce a specific structure-aware loss to encourage the NMT model to better learn the structure knowledge. Experimental results on WMT17, WMT18 and WMT19 Chinese-to-English translation tasks demonstrate the effectiveness of the proposed methods.

Neural machine translation (MT) models obtain state-of-the-art performance while maintaining a simple, end-to-end architecture. However, little is known about what these models learn about source and target languages during the training process.

Adapter layers are lightweight, learnable units inserted between transformer layers. Recent work explores using such layers for neural machine translation (NMT), to adapt pre-trained models to new domains or language pairs, training only a small set of parameters for each new setting (language pair or domain). In this work we study the compositionality of language and domain adapters in the context of Machine Translation. We aim to study, 1) parameter-efficient adaptation to multiple domains and languages simultaneously (full-resource scenario) and 2) cross-lingual transfer in domains where parallel data is unavailable for certain language pairs (partial-resource scenario). We find that in the partial resource scenario a naive combination of domain-specific and language-specific adapters often results in `catastrophic forgetting' of the missing languages. We study other ways to combine the adapters to alleviate this issue and maximize cross-lingual transfer. With our best adapter combinations, we obtain improvements of 3-4 BLEU on average for source languages that do not have in-domain data. For target languages without in-domain data, we achieve a similar improvement by combining adapters with back-translation. Supplementary material is available at https://tinyurl.com/r66stbxj.

Language reconciles the ideas, beliefs and values of people from diverse cultural, social, economic, religious and professional backgrounds, and it is imperative for sustainable development. Undoubtedly, the emergence of Neural Machine Translation has gained significant advancement in language translation automation, and consequently, establishing seamless reconciliation between two communicating entities with diverse backgrounds. Comprising of sequence to sequence architecture, Neural Machine Translation models have outperformed statistical and rule-based models in terms of concordance and syntax, regardless of how complex the language structure is. However, structured language resources are scarce for low-resource languages like Akan. Research contributions toward the revitalization of these languages are limited, along with subtle adverse events such as cultural assimilation and language imperialism. In order to solve the problem of low-resource machine translation from Akan to English, we mine and use the first parallel corpus for the Akan-English translation. We establish a benchmark for the Akan-English translation using a deep hierarchical end-to-end attention-based neural machine translation model trained on the parallel corpus. Experimental results further confirm the effectiveness of our approach towards the task. Nevertheless, a confirmed grammatical parity between the two languages bequests improvement approaching a substantial revitalization of the Akan language.

Generative adversarial networks (GANs) has achieved great success in the field of image processing, Adversarial Neural Machine Translation(NMT) is the application of GANs to machine translation. Unlike previous work training NMT model through maximizing the likelihood of the human translation, Adversarial NMT minimizes the distinction between human translation and the translation generated by a NMT model. Even though Adversarial NMT has achieved impressive results, while using little in the way of prior knowledge. In this paper, we integrated bilingual dictionaries to Adversarial NMT by leveraging a character model. Extensive experiment shows that our proposed methods can achieve remarkable improvement on the translation quality of Adversarial NMT, and obtain better result than several strong baselines.

In this paper we describe our team's (NICT-5) Neural Machine Translation (NMT) models whose translations were submitted to shared tasks of the 7 th Workshop on Asian Translation.We participated in the Indic language multilingual sub-task as well as the NICT-SAP multilingual multi-domain sub-task.We focused on naive many-to-many NMT models which gave reasonable translation quality despite their simplicity.Our observations are twofold: (a.) Many-to-many models suffer from a lack of consistency where the translation quality for some language pairs is very good but for some others it is terrible when compared against one-to-many and many-toone baselines.(b.) Oversampling smaller corpora does not necessarily give the best translation quality for the language pair associated with that pair.