We investigate two specific manifestations of compositionality in Neural Machine Translation (NMT) : (1) Productivity - the ability of the model to extend its predictions beyond the observed length in training data and (2) Systematicity - the ability of the model to systematically recombine known parts and rules. We evaluate a standard Sequence to Sequence model on tests designed to assess these two properties in NMT. We quantitatively demonstrate that inadequate temporal processing, in the form of poor encoder representations is a bottleneck for both Productivity and Systematicity. We propose a simple pre-training mechanism which alleviates model performance on the two properties and leads to a significant improvement in BLEU scores.
We introduce Data Diversification: a simple but effective strategy to boost neural machine translation (NMT) performance. It diversifies the training data by using the predictions of multiple forward and backward models and then merging them with the original dataset on which the final NMT model is trained. Our method is applicable to all NMT models. It does not require extra monolingual data like back-translation, nor does it add more computations and parameters like ensembles of models. Our method achieves state-of-the-art BLEU scores of 30.7 and 43.7 in the WMT'14 English-German and English-French translation tasks, respectively. It also substantially improves on 8 other translation tasks: 4 IWSLT tasks (English-German and English-French) and 4 low-resource translation tasks (English-Nepali and English-Sinhala). We demonstrate that our method is more effective than knowledge distillation and dual learning, it exhibits strong correlation with ensembles of models, and it trades perplexity off for better BLEU score. We have released our source code at https://github.com/nxphi47/data_diversification
We implement a Tensor Train layer in the TensorFlow Neural Machine Translation (NMT) model using the t3f library. We perform training runs on the IWSLT English-Vietnamese '15 and WMT German-English '16 datasets with learning rates $\in \{0.0004,0.0008,0.0012\}$, maximum ranks $\in \{2,4,8,16\}$ and a range of core dimensions. We compare against a target BLEU test score of 24.0, obtained by our benchmark run. For the IWSLT English-Vietnamese training, we obtain BLEU test/dev scores of 24.0/21.9 and 24.2/21.9 using core dimensions $(2, 2, 256) \times (2, 2, 512)$ with learning rate 0.0012 and rank distributions $(1,4,4,1)$ and $(1,4,16,1)$ respectively. These runs use 113\% and 397\% of the flops of the benchmark run respectively. We find that, of the parameters surveyed, a higher learning rate and more `rectangular' core dimensions generally produce higher BLEU scores. For the WMT German-English dataset, we obtain BLEU scores of 24.0/23.8 using core dimensions $(4, 4, 128) \times (4, 4, 256)$ with learning rate 0.0012 and rank distribution $(1,2,2,1)$. We discuss the potential for future optimization and application of Tensor Train decomposition to other NMT models.
In this paper, we present a conditional GAN with two generators and a common discriminator for multiview learning problems where observations have two views, but one of them may be missing for some of the training samples. This is for example the case for multilingual collections where documents are not available in all languages. Some studies tackled this problem by assuming the existence of view generation functions to approximately complete the missing views; for example Machine Translation to translate documents into the missing languages. These functions generally require an external resource to be set and their quality has a direct impact on the performance of the learned multiview classifier over the completed training set. Our proposed approach addresses this problem by jointly learning the missing views and the multiview classifier using a tripartite game with two generators and a discriminator. Each of the generators is associated to one of the views and tries to fool the discriminator by generating the other missing view conditionally on the corresponding observed view. The discriminator then tries to identify if for an observation, one of its views is completed by one of the generators or if both views are completed along with its class. Our results on a subset of Reuters RCV1/RCV2 collections show that the discriminator achieves significant classification performance; and that the generators learn the missing views with high quality without the need of any consequent external resource.
In this paper, we identify an interesting kind of error in the output of Unsupervised Neural Machine Translation (UNMT) systems like \textit{Undreamt}(footnote). We refer to this error type as \textit{Scrambled Translation problem}. We observe that UNMT models which use \textit{word shuffle} noise (as in case of Undreamt) can generate correct words, but fail to stitch them together to form phrases. As a result, words of the translated sentence look \textit{scrambled}, resulting in decreased BLEU. We hypothesise that the reason behind \textit{scrambled translation problem} is 'shuffling noise' which is introduced in every input sentence as a denoising strategy. To test our hypothesis, we experiment by retraining UNMT models with a simple \textit{retraining} strategy. We stop the training of the Denoising UNMT model after a pre-decided number of iterations and resume the training for the remaining iterations -- which number is also pre-decided -- using original sentence as input without adding any noise. Our proposed solution achieves significant performance improvement UNMT models that train conventionally. We demonstrate these performance gains on four language pairs, \textit{viz.}, English-French, English-German, English-Spanish, Hindi-Punjabi. Our qualitative and quantitative analysis shows that the retraining strategy helps achieve better alignment as observed by attention heatmap and better phrasal translation, leading to statistically significant improvement in BLEU scores.
The research in machine translation community focus on translation in text space. However, humans are in fact also good at direct translation in pronunciation space. Some existing translation systems, such as simultaneous machine translation, are inherently more natural and thus potentially more robust by directly translating in pronunciation space. In this paper, we conduct large scale experiments on a self-built dataset with about $20$M En-Zh pairs of text sentences and corresponding pronunciation sentences. We proposed three new categories of translations: $1)$ translating a pronunciation sentence in source language into a pronunciation sentence in target language (P2P-Tran), $2)$ translating a text sentence in source language into a pronunciation sentence in target language (T2P-Tran), and $3)$ translating a pronunciation sentence in source language into a text sentence in target language (P2T-Tran), and compare them with traditional text translation (T2T-Tran). Our experiments clearly show that all $4$ categories of translations have comparable performances, with small and sometimes ignorable differences.
This work introduces a machine translation task where the output is aimed at audiences of different levels of target language proficiency. We collect a high quality dataset of news articles available in English and Spanish, written for diverse grade levels and propose a method to align segments across comparable bilingual articles. The resulting dataset makes it possible to train multi-task sequence-to-sequence models that translate Spanish into English targeted at an easier reading grade level than the original Spanish. We show that these multi-task models outperform pipeline approaches that translate and simplify text independently.
In this paper, we propose a new metric for Machine Translation (MT) evaluation, based on bi-directional entailment. We show that machine generated translation can be evaluated by determining paraphrasing with a reference translation provided by a human translator. We hypothesize, and show through experiments, that paraphrasing can be detected by evaluating entailment relationship in the forward and backward direction. Unlike conventional metrics, like BLEU or METEOR, our approach uses deep learning to determine the semantic similarity between candidate and reference translation for generating scores rather than relying upon simple n-gram overlap. We use BERT's pre-trained implementation of transformer networks, fine-tuned on MNLI corpus, for natural language inferencing. We apply our evaluation metric on WMT'14 and WMT'17 dataset to evaluate systems participating in the translation task and find that our metric has a better correlation with the human annotated score compared to the other traditional metrics at system level.
We analyse coreference phenomena in three neural machine translation systems trained with different data settings with or without access to explicit intra- and cross-sentential anaphoric information. We compare system performance on two different genres: news and TED talks. To do this, we manually annotate (the possibly incorrect) coreference chains in the MT outputs and evaluate the coreference chain translations. We define an error typology that aims to go further than pronoun translation adequacy and includes types such as incorrect word selection or missing words. The features of coreference chains in automatic translations are also compared to those of the source texts and human translations. The analysis shows stronger potential translationese effects in machine translated outputs than in human translations.
The dominant approach to sequence generation is to produce a sequence in some predefined order, e.g. left to right. In contrast, we propose a more general model that can generate the output sequence by inserting tokens in any arbitrary order. Our model learns decoding order as a result of its training procedure. Our experiments show that this model is superior to fixed order models on a number of sequence generation tasks, such as Machine Translation, Image-to-LaTeX and Image Captioning.
Adaptive Precision Training: Quantify Back Propagation in Neural Networks with Fixed-point Numbers. Recent emerged quantization technique has been applied to inference of deep neural networks for fast and efficient execution. However, directly applying quantization in training can cause significant accuracy loss, thus remaining an open challenge.
Despite the recent success of deep neural networks in natural language processing (NLP), their interpretability remains a challenge. We analyze the representations learned by neural machine translation models at various levels of granularity and evaluate their quality through relevant extrinsic properties. In particular, we seek answers to the following questions: (i) How accurately is word-structure captured within the learned representations, an important aspect in translating morphologically-rich languages? (ii) Do the representations capture long-range dependencies, and effectively handle syntactically divergent languages? (iii) Do the representations capture lexical semantics? We conduct a thorough investigation along several parameters: (i) Which layers in the architecture capture each of these linguistic phenomena; (ii) How does the choice of translation unit (word, character, or subword unit) impact the linguistic properties captured by the underlying representations? (iii) Do the encoder and decoder learn differently and independently? (iv) Do the representations learned by multilingual NMT models capture the same amount of linguistic information as their bilingual counterparts? Our data-driven, quantitative evaluation illuminates important aspects in NMT models and their ability to capture various linguistic phenomena. We show that deep NMT models learn a non-trivial amount of linguistic information. Notable findings include: i) Word morphology and part-of-speech information are captured at the lower layers of the model; (ii) In contrast, lexical semantics or non-local syntactic and semantic dependencies are better represented at the higher layers; (iii) Representations learned using characters are more informed about wordmorphology compared to those learned using subword units; and (iv) Representations learned by multilingual models are richer compared to bilingual models.
While deep learning is a powerful tool for natural language processing (NLP) problems, successful solutions to these problems rely heavily on large amounts of annotated samples. However, manually annotating data is expensive and time-consuming. Active Learning (AL) strategies reduce the need for huge volumes of labeled data by iteratively selecting a small number of examples for manual annotation based on their estimated utility in training the given model. In this paper, we argue that since AL strategies choose examples independently, they may potentially select similar examples, all of which may not contribute significantly to the learning process. Our proposed approach, Active$\mathbf{^2}$ Learning (A$\mathbf{^2}$L), actively adapts to the deep learning model being trained to eliminate such redundant examples chosen by an AL strategy. We show that A$\mathbf{^2}$L is widely applicable by using it in conjunction with several different AL strategies and NLP tasks. We empirically demonstrate that the proposed approach is further able to reduce the data requirements of state-of-the-art AL strategies by $\approx \mathbf{3-25\%}$ on an absolute scale on multiple NLP tasks while achieving the same performance with virtually no additional computation overhead.
Most neural machine translation systems still translate sentences in isolation. To make further progress, a promising line of research additionally considers the surrounding context in order to provide the model potentially missing source-side information, as well as to maintain a coherent output. One difficulty in training such larger-context (i.e. document-level) machine translation systems is that context may be missing from many parallel examples. To circumvent this issue, two-stage approaches, in which sentence-level translations are post-edited in context, have recently been proposed. In this paper, we instead consider the viability of filling in the missing context. In particular, we consider three distinct approaches to generate the missing context: using random contexts, applying a copy heuristic or generating it with a language model. In particular, the copy heuristic significantly helps with lexical coherence, while using completely random contexts hurts performance on many long-distance linguistic phenomena. We also validate the usefulness of tagged back-translation. In addition to improving BLEU scores as expected, using back-translated data helps larger-context machine translation systems to better capture long-range phenomena.
Recently, neural models led to significant improvements in both machine translation (MT) and natural language generation tasks (NLG). However, generation of long descriptive summaries conditioned on structured data remains an open challenge. Likewise, MT that goes beyond sentence-level context is still an open issue (e.g., document-level MT or MT with metadata). To address these challenges, we propose to leverage data from both tasks and do transfer learning between MT, NLG, and MT with source-side metadata (MT+NLG). First, we train document-based MT systems with large amounts of parallel data. Then, we adapt these models to pure NLG and MT+NLG tasks by fine-tuning with smaller amounts of domain-specific data. This end-to-end NLG approach, without data selection and planning, outperforms the previous state of the art on the Rotowire NLG task. We participated to the "Document Generation and Translation" task at WNGT 2019, and ranked first in all tracks.
Standard neural machine translation (NMT) is on the assumption that the document-level context is independent. Most existing document-level NMT approaches are satisfied with a smattering sense of global document-level information, while this work focuses on exploiting detailed document-level context in terms of a memory network. The capacity of the memory network that detecting the most relevant part of the current sentence from memory renders a natural solution to model the rich document-level context. In this work, the proposed document-aware memory network is implemented to enhance the Transformer NMT baseline. Experiments on several tasks show that the proposed method significantly improves the NMT performance over strong Transformer baselines and other related studies.
We share a French-English parallel corpus of Foursquare restaurant reviews (https://europe.naverlabs.com/research/natural-language-processing/machine-translation-of-restaurant-reviews), and define a new task to encourage research on Neural Machine Translation robustness and domain adaptation, in a real-world scenario where better-quality MT would be greatly beneficial. We discuss the challenges of such user-generated content, and train good baseline models that build upon the latest techniques for MT robustness. We also perform an extensive evaluation (automatic and human) that shows significant improvements over existing online systems. Finally, we propose task-specific metrics based on sentiment analysis or translation accuracy of domain-specific polysemous words.
Multilingual Neural Machine Translation (MNMT) for low-resource languages (LRL) can be enhanced by the presence of related high-resource languages (HRL), but the relatedness of HRL usually relies on predefined linguistic assumptions about language similarity. Recently, adapting MNMT to a LRL has shown to greatly improve performance. In this work, we explore the problem of adapting an MNMT model to an unseen LRL using data selection and model adaptation. In order to improve NMT for LRL, we employ perplexity to select HRL data that are most similar to the LRL on the basis of language distance. We extensively explore data selection in popular multilingual NMT settings, namely in (zero-shot) translation, and in adaptation from a multilingual pre-trained model, for both directions (LRL-en). We further show that dynamic adaptation of the model's vocabulary results in a more favourable segmentation for the LRL in comparison with direct adaptation. Experiments show reductions in training time and significant performance gains over LRL baselines, even with zero LRL data (+13.0 BLEU), up to +17.0 BLEU for pre-trained multilingual model dynamic adaptation with related data selection. Our method outperforms current approaches, such as massively multilingual models and data augmentation, on four LRL.
Translation into morphologically-rich languages challenges neural machine translation (NMT) models with extremely sparse vocabularies where atomic treatment of surface forms is unrealistic. This problem is typically addressed by either pre-processing words into subword units or performing translation directly at the level of characters. The former is based on word segmentation algorithms optimized using corpus-level statistics with no regard to the translation task. The latter learns directly from translation data but requires rather deep architectures. In this paper, we propose to translate words by modeling word formation through a hierarchical latent variable model which mimics the process of morphological inflection. Our model generates words one character at a time by composing two latent representations: a continuous one, aimed at capturing the lexical semantics, and a set of (approximately) discrete features, aimed at capturing the morphosyntactic function, which are shared among different surface forms. Our model achieves better accuracy in translation into three morphologically-rich languages than conventional open-vocabulary NMT methods, while also demonstrating a better generalization capacity under low to mid-resource settings.
We present BART, a denoising autoencoder for pretraining sequence-to-sequence models. BART is trained by (1) corrupting text with an arbitrary noising function, and (2) learning a model to reconstruct the original text. It uses a standard Tranformer-based neural machine translation architecture which, despite its simplicity, can be seen as generalizing BERT (due to the bidirectional encoder), GPT (with the left-to-right decoder), and many other more recent pretraining schemes. We evaluate a number of noising approaches, finding the best performance by both randomly shuffling the order of the original sentences and using a novel in-filling scheme, where spans of text are replaced with a single mask token. BART is particularly effective when fine tuned for text generation but also works well for comprehension tasks. It matches the performance of RoBERTa with comparable training resources on GLUE and SQuAD, achieves new state-of-the-art results on a range of abstractive dialogue, question answering, and summarization tasks, with gains of up to 6 ROUGE. BART also provides a 1.1 BLEU increase over a back-translation system for machine translation, with only target language pretraining. We also report ablation experiments that replicate other pretraining schemes within the BART framework, to better measure which factors most influence end-task performance.