PyTorchKR Papers

최신 논문

총 5,082편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2104.04497공식 코드 · Papers with Code

Chinese Character Decomposition for Neural MT with Multi-Word Expressions

Chinese character decomposition has been used as a feature to enhance Machine Translation (MT) models, combining radicals into character and word level models. Recent work has investigated ideograph or stroke level embedding. However, questions remain about different decomposition levels of Chinese character representations, radical and strokes, best suited for MT. To investigate the impact of Chinese decomposition embedding in detail, i.e., radical, stroke, and intermediate levels, and how well these decompositions represent the meaning of the original character sequences, we carry out analysis with both automated and human evaluation of MT. Furthermore, we investigate if the combination of decomposed Multiword Expressions (MWEs) can enhance the model learning. MWE integration into MT has seen more than a decade of exploration. However, decomposed MWEs has not previously been explored.

arXiv:2104.04125

Design and Implementation of English To Yorùbá Verb Phrase Machine Translation System

We aim to develop an English-to-Yoruba machine translation system which can translate English verb phrase text to its Yoruba equivalent.Words from both languages Source Language and Target Language were collected for the verb phrase group in the home domain. The lexical translation is done by assigning values of the matching word in the dictionary. The syntax of the two languages was realized using Context-Free Grammar, we validated the rewrite rules with finite state automata. The human evaluation method was used and expert fluency was scored. The evaluation shows the system performed better than that of sampled Google translation with over 70 percent of the response matching that of the system's output.

arXiv:2104.04087공식 코드 · Papers with Code

A Sketch-Based Neural Model for Generating Commit Messages from Diffs

Commit messages have an important impact in software development, especially when working in large teams. Multiple developers who have a different style of writing may often be involved in the same project. For this reason, it may be difficult to maintain a strict pattern of writing informative commit messages, with the most frequent issue being that these messages are not descriptive enough. In this paper we apply neural machine translation (NMT) techniques to convert code diffs into commit messages and we present an improved sketch-based encoder for this task. We split the approach into three parts. Firstly, we focus on finding a more suitable NMT baseline for this problem. Secondly, we show that the performance of the NMT models can be improved by training on examples containing a specific file type. Lastly, we introduce a novel sketch-based neural model inspired by recent approaches used for code generation and we show that the sketch-based encoder significantly outperforms existing state of the art solutions. The results highlight that this improvement is relevant especially for Java source code files, by examining two different datasets introduced in recent years for this task.

arXiv:2104.03391

Interpreting Verbal Metaphors by Paraphrasing

Metaphorical expressions are difficult linguistic phenomena, challenging diverse Natural Language Processing tasks. Previous works showed that paraphrasing a metaphor as its literal counterpart can help machines better process metaphors on downstream tasks. In this paper, we interpret metaphors with BERT and WordNet hypernyms and synonyms in an unsupervised manner, showing that our method significantly outperforms the state-of-the-art baseline. We also demonstrate that our method can help a machine translation system improve its accuracy in translating English metaphors to 8 target languages.

arXiv:2104.03958공식 코드 · Papers with Code

GrASP: A Library for Extracting and Exploring Human-Interpretable Textual Patterns

Data exploration is an important step of every data science and machine learning project, including those involving textual data. We provide a novel language tool, in the form of a publicly available Python library for extracting patterns from textual data. The library integrates a first public implementation of the existing GrASP algorithm. It allows users to extract patterns using a number of general-purpose built-in linguistic attributes (such as hypernyms, part-of-speech tags, and syntactic dependency tags), as envisaged for the original algorithm, as well as domain-specific custom attributes which can be incorporated into the library by implementing two functions. The library is equipped with a web-based interface empowering human users to conveniently explore data via the extracted patterns, using complementary pattern-centric and example-centric views: the former includes a reading in natural language and statistics of each extracted pattern; the latter shows applications of each extracted pattern to training examples. We demonstrate the usefulness of the library in classification (spam detection and argument mining), model analysis (machine translation), and artifact discovery in datasets (SNLI and 20Newsgroups).

arXiv:2104.03543

Extended Parallel Corpus for Amharic-English Machine Translation

This paper describes the acquisition, preprocessing, segmentation, and alignment of an Amharic-English parallel corpus. It will be helpful for machine translation of a low-resource language, Amharic. We freely released the corpus for research purposes. Furthermore, we developed baseline statistical and neural machine translation systems; we trained statistical and neural machine translation models using the corpus. In the experiments, we also used a large monolingual corpus for the language model of statistical machine translation and back-translation of neural machine translation. In the automatic evaluation, neural machine translation models outperform statistical machine translation models by approximately six to seven Bilingual Evaluation Understudy (BLEU) points. Besides, among the neural machine translation models, the subword models outperform the word-based models by three to four BLEU points. Moreover, two other relevant automatic evaluation metrics, Translation Edit Rate on Character Level and Better Evaluation as Ranking, reflect corresponding differences among the trained models.

arXiv:2104.02831공식 코드 · Papers with Code

Better Neural Machine Translation by Extracting Linguistic Information from BERT

Adding linguistic information (syntax or semantics) to neural machine translation (NMT) has mostly focused on using point estimates from pre-trained models. Directly using the capacity of massive pre-trained contextual word embedding models such as BERT (Devlin et al., 2019) has been marginally useful in NMT because effective fine-tuning is difficult to obtain for NMT without making training brittle and unreliable. We augment NMT by extracting dense fine-tuned vector-based linguistic information from BERT instead of using point estimates. Experimental results show that our method of incorporating linguistic information helps NMT to generalize better in a variety of training contexts and is no more difficult to train than conventional Transformer-based NMT.

arXiv:2104.02858

Capturing Multi-Resolution Context by Dilated Self-Attention

Self-attention has become an important and widely used neural network component that helped to establish new state-of-the-art results for various applications, such as machine translation and automatic speech recognition (ASR). However, the computational complexity of self-attention grows quadratically with the input sequence length. This can be particularly problematic for applications such as ASR, where an input sequence generated from an utterance can be relatively long. In this work, we propose a combination of restricted self-attention and a dilation mechanism, which we refer to as dilated self-attention. The restricted self-attention allows attention to neighboring frames of the query at a high resolution, and the dilation mechanism summarizes distant information to allow attending to it with a lower resolution. Different methods for summarizing distant frames are studied, such as subsampling, mean-pooling, and attention-based pooling. ASR results demonstrate substantial improvements compared to restricted self-attention alone, achieving similar results compared to full-sequence based self-attention with a fraction of the computational costs.

arXiv:2104.02516

AI4D -- African Language Program

Advances in speech and language technologies enable tools such as voice-search, text-to-speech, speech recognition and machine translation. These are however only available for high resource languages like English, French or Chinese. Without foundational digital resources for African languages, which are considered low-resource in the digital context, these advanced tools remain out of reach. This work details the AI4D - African Language Program, a 3-part project that 1) incentivised the crowd-sourcing, collection and curation of language datasets through an online quantitative and qualitative challenge, 2) supported research fellows for a period of 3-4 months to create datasets annotated for NLP tasks, and 3) hosted competitive Machine Learning challenges on the basis of these datasets. Key outcomes of the work so far include 1) the creation of 9+ open source, African language datasets annotated for a variety of ML tasks, and 2) the creation of baseline models for these datasets through hosting of competitive ML challenges.

arXiv:2104.02308

ODE Transformer: An Ordinary Differential Equation-Inspired Model for Neural Machine Translation

It has been found that residual networks are an Euler discretization of solutions to Ordinary Differential Equations (ODEs). In this paper, we explore a deeper relationship between Transformer and numerical methods of ODEs. We show that a residual block of layers in Transformer can be described as a higher-order solution to ODEs. This leads us to design a new architecture (call it ODE Transformer) analogous to the Runge-Kutta method that is well motivated in ODEs. As a natural extension to Transformer, ODE Transformer is easy to implement and parameter efficient. Our experiments on three WMT tasks demonstrate the genericity of this model, and large improvements in performance over several strong baselines. It achieves 30.76 and 44.11 BLEU scores on the WMT'14 En-De and En-Fr test data. This sets a new state-of-the-art on the WMT'14 En-Fr task.

arXiv:2104.01853공식 코드 · Papers with Code

Rethinking Perturbations in Encoder-Decoders for Fast Training

We often use perturbations to regularize neural models. For neural encoder-decoders, previous studies applied the scheduled sampling (Bengio et al., 2015) and adversarial perturbations (Sato et al., 2019) as perturbations but these methods require considerable computational time. Thus, this study addresses the question of whether these approaches are efficient enough for training time. We compare several perturbations in sequence-to-sequence problems with respect to computational time. Experimental results show that the simple techniques such as word dropout (Gal and Ghahramani, 2016) and random replacement of input tokens achieve comparable (or better) scores to the recently proposed perturbations, even though these simple methods are faster. Our code is publicly available at https://github.com/takase/rethink_perturbations.

arXiv:2104.01264공식 코드 · Papers with Code

Attention Forcing for Machine Translation

Auto-regressive sequence-to-sequence models with attention mechanisms have achieved state-of-the-art performance in various tasks including Text-To-Speech (TTS) and Neural Machine Translation (NMT). The standard training approach, teacher forcing, guides a model with the reference output history. At inference stage, the generated output history must be used. This mismatch can impact performance. However, it is highly challenging to train the model using the generated output. Several approaches have been proposed to address this problem, normally by selectively using the generated output history. To make training stable, these approaches often require a heuristic schedule or an auxiliary classifier. This paper introduces attention forcing for NMT. This approach guides the model with the generated output history and reference attention, and can reduce the training-inference mismatch without a schedule or a classifier. Attention forcing has been successful in TTS, but its application to NMT is more challenging, due to the discrete and multi-modal nature of the output space. To tackle this problem, this paper adds a selection scheme to vanilla attention forcing, which automatically selects a suitable training approach for each pair of training data. Experiments show that attention forcing can improve the overall translation quality and the diversity of the translations.

arXiv:2104.00041

Domain-specific MT for Low-resource Languages: The case of Bambara-French

Translating to and from low-resource languages is a challenge for machine translation (MT) systems due to a lack of parallel data. In this paper we address the issue of domain-specific MT for Bambara, an under-resourced Mande language spoken in Mali. We present the first domain-specific parallel dataset for MT of Bambara into and from French. We discuss challenges in working with small quantities of domain-specific data for a low-resource language and we present the results of machine learning experiments on this data.

arXiv:2104.00664공식 코드 · Papers with Code

Sampling and Filtering of Neural Machine Translation Distillation Data

In most of neural machine translation distillation or stealing scenarios, the goal is to preserve the performance of the target model (teacher). The highest-scoring hypothesis of the teacher model is commonly used to train a new model (student). If reference translations are also available, then better hypotheses (with respect to the references) can be upsampled and poor hypotheses either removed or undersampled. This paper explores the importance sampling method landscape (pruning, hypothesis upsampling and undersampling, deduplication and their combination) with English to Czech and English to German MT models using standard MT evaluation metrics. We show that careful upsampling and combination with the original data leads to better performance when compared to training only on the original or synthesized data or their direct combination.

arXiv:2104.00366

Low-Resource Neural Machine Translation for Southern African Languages

Low-resource African languages have not fully benefited from the progress in neural machine translation because of a lack of data. Motivated by this challenge we compare zero-shot learning, transfer learning and multilingual learning on three Bantu languages (Shona, isiXhosa and isiZulu) and English. Our main target is English-to-isiZulu translation for which we have just 30,000 sentence pairs, 28% of the average size of our other corpora. We show the importance of language similarity on the performance of English-to-isiZulu transfer learning based on English-to-isiXhosa and English-to-Shona parent models whose BLEU scores differ by 5.2. We then demonstrate that multilingual learning surpasses both transfer learning and zero-shot learning on our dataset, with BLEU score improvements relative to the baseline English-to-isiZulu model of 9.9, 6.1 and 2.0 respectively. Our best model also improves the previous SOTA BLEU score by more than 10.

arXiv:2104.00267

Detecting over/under-translation errors for determining adequacy in human translations

We present a novel approach to detecting over and under translations (OT/UT) as part of adequacy error checks in translation evaluation. We do not restrict ourselves to machine translation (MT) outputs and specifically target applications with human generated translation pipeline. The goal of our system is to identify OT/UT errors from human translated video subtitles with high error recall. We achieve this without reference translations by learning a model on synthesized training data. We compare various classification networks that we trained on embeddings from pre-trained language model with our best hybrid network of GRU + CNN achieving 89.3% accuracy on high-quality human-annotated evaluation data in 8 languages.

arXiv:2104.00676

Is Label Smoothing Truly Incompatible with Knowledge Distillation: An Empirical Study

This work aims to empirically clarify a recently discovered perspective that label smoothing is incompatible with knowledge distillation. We begin by introducing the motivation behind on how this incompatibility is raised, i.e., label smoothing erases relative information between teacher logits. We provide a novel connection on how label smoothing affects distributions of semantically similar and dissimilar classes. Then we propose a metric to quantitatively measure the degree of erased information in sample's representation. After that, we study its one-sidedness and imperfection of the incompatibility view through massive analyses, visualizations and comprehensive experiments on Image Classification, Binary Networks, and Neural Machine Translation. Finally, we broadly discuss several circumstances wherein label smoothing will indeed lose its effectiveness. Project page: http://zhiqiangshen.com/projects/LS_and_KD/index.html.

arXiv:2104.00332

UC2: Universal Cross-lingual Cross-modal Vision-and-Language Pre-training

Vision-and-language pre-training has achieved impressive success in learning multimodal representations between vision and language. To generalize this success to non-English languages, we introduce UC2, the first machine translation-augmented framework for cross-lingual cross-modal representation learning. To tackle the scarcity problem of multilingual captions for image datasets, we first augment existing English-only datasets with other languages via machine translation (MT). Then we extend the standard Masked Language Modeling and Image-Text Matching training objectives to multilingual setting, where alignment between different languages is captured through shared visual context (i.e, using image as pivot). To facilitate the learning of a joint embedding space of images and all languages of interest, we further propose two novel pre-training tasks, namely Masked Region-to-Token Modeling (MRTM) and Visual Translation Language Modeling (VTLM), leveraging MT-enhanced translated data. Evaluation on multilingual image-text retrieval and multilingual visual question answering benchmarks demonstrates that our proposed framework achieves new state-of-the-art on diverse non-English benchmarks while maintaining comparable performance to monolingual pre-trained models on English tasks.

arXiv:2104.00290공식 코드 · Papers with Code

Many-to-English Machine Translation Tools, Data, and Pretrained Models

While there are more than 7000 languages in the world, most translation research efforts have targeted a few high-resource languages. Commercial translation systems support only one hundred languages or fewer, and do not make these models available for transfer to low resource languages. In this work, we present useful tools for machine translation research: MTData, NLCodec, and RTG. We demonstrate their usefulness by creating a multilingual neural machine translation model capable of translating from 500 source languages to English. We make this multilingual model readily downloadable and usable as a service, or as a parent model for transfer-learning to even lower-resource languages.

arXiv:2104.00106공식 코드 · Papers with Code

Zero-Shot Language Transfer vs Iterative Back Translation for Unsupervised Machine Translation

This work focuses on comparing different solutions for machine translation on low resource language pairs, namely, with zero-shot transfer learning and unsupervised machine translation. We discuss how the data size affects the performance of both unsupervised MT and transfer learning. Additionally we also look at how the domain of the data affects the result of unsupervised MT. The code to all the experiments performed in this project are accessible on Github.