PyTorchKR Papers

최신 논문

총 1,781편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:1908.05672공식 코드 · Papers with Code

Towards Making the Most of BERT in Neural Machine Translation

GPT-2 and BERT demonstrate the effectiveness of using pre-trained language models (LMs) on various natural language processing tasks. However, LM fine-tuning often suffers from catastrophic forgetting when applied to resource-rich tasks. In this work, we introduce a concerted training framework (CTNMT) that is the key to integrate the pre-trained LMs to neural machine translation (NMT). Our proposed CTNMT consists of three techniques: a) asymptotic distillation to ensure that the NMT model can retain the previous pre-trained knowledge; b) a dynamic switching gate to avoid catastrophic forgetting of pre-trained knowledge; and c) a strategy to adjust the learning paces according to a scheduled policy. Our experiments in machine translation show CTNMT gains of up to 3 BLEU score on the WMT14 English-German language pair which even surpasses the previous state-of-the-art pre-training aided NMT by 1.4 BLEU score. While for the large WMT14 English-French task with 40 millions of sentence-pairs, our base model still significantly improves upon the state-of-the-art Transformer big model by more than 1 BLEU score. The code and model can be downloaded from https://github.com/bytedance/neurst/ tree/master/examples/ctnmt.

arXiv:1908.05957공식 코드 · Papers with Code

Densely Connected Graph Convolutional Networks for Graph-to-Sequence Learning

We focus on graph-to-sequence learning, which can be framed as transducing graph structures to sequences for text generation. To capture structural information associated with graphs, we investigate the problem of encoding graphs using graph convolutional networks (GCNs). Unlike various existing approaches where shallow architectures were used for capturing local structural information only, we introduce a dense connection strategy, proposing a novel Densely Connected Graph Convolutional Networks (DCGCNs). Such a deep architecture is able to integrate both local and non-local features to learn a better structural representation of a graph. Our model outperforms the state-of-the-art neural models significantly on AMRto-text generation and syntax-based neural machine translation.

arXiv:1908.05204공식 코드 · Papers with Code

On The Evaluation of Machine Translation Systems Trained With Back-Translation

Back-translation is a widely used data augmentation technique which leverages target monolingual data. However, its effectiveness has been challenged since automatic metrics such as BLEU only show significant improvements for test examples where the source itself is a translation, or translationese. This is believed to be due to translationese inputs better matching the back-translated training data. In this work, we show that this conjecture is not empirically supported and that back-translation improves translation quality of both naturally occurring text as well as translationese according to professional human translators. We provide empirical evidence to support the view that back-translation is preferred by humans because it produces more fluent outputs. BLEU cannot capture human preferences because references are translationese when source sentences are natural text. We recommend complementing BLEU with a language model score to measure fluency.

arXiv:1908.03265공식 코드 · Papers with Code

On the Variance of the Adaptive Learning Rate and Beyond

The learning rate warmup heuristic achieves remarkable success in stabilizing training, accelerating convergence and improving generalization for adaptive stochastic optimization algorithms like RMSprop and Adam. Here, we study its mechanism in details. Pursuing the theory behind warmup, we identify a problem of the adaptive learning rate (i.e., it has problematically large variance in the early stage), suggest warmup works as a variance reduction technique, and provide both empirical and theoretical evidence to verify our hypothesis. We further propose RAdam, a new variant of Adam, by introducing a term to rectify the variance of the adaptive learning rate. Extensive experimental results on image classification, language modeling, and neural machine translation verify our intuition and demonstrate the effectiveness and robustness of our proposed method. All implementations are available at: https://github.com/LiyuanLucasLiu/RAdam.

arXiv:1908.00681공식 코드 · Papers with Code

FlowSense: A Natural Language Interface for Visual Data Exploration within a Dataflow System

Dataflow visualization systems enable flexible visual data exploration by allowing the user to construct a dataflow diagram that composes query and visualization modules to specify system functionality. However learning dataflow diagram usage presents overhead that often discourages the user. In this work we design FlowSense, a natural language interface for dataflow visualization systems that utilizes state-of-the-art natural language processing techniques to assist dataflow diagram construction. FlowSense employs a semantic parser with special utterance tagging and special utterance placeholders to generalize to different datasets and dataflow diagrams. It explicitly presents recognized dataset and diagram special utterances to the user for dataflow context awareness. With FlowSense the user can expand and adjust dataflow diagrams more conveniently via plain English. We apply FlowSense to the VisFlow subset-flow visualization system to enhance its usability. We evaluate FlowSense by one case study with domain experts on a real-world data analysis problem and a formal user study.

arXiv:1907.12484공식 코드 · Papers with Code

Joey NMT: A Minimalist NMT Toolkit for Novices

We present Joey NMT, a minimalist neural machine translation toolkit based on PyTorch that is specifically designed for novices. Joey NMT provides many popular NMT features in a small and simple code base, so that novices can easily and quickly learn to use it and adapt it to their needs. Despite its focus on simplicity, Joey NMT supports classic architectures (RNNs, transformers), fast beam search, weight tying, and more, and achieves performance comparable to more complex toolkits on standard benchmarks. We evaluate the accessibility of our toolkit in a user study where novices with general knowledge about Pytorch and NMT and experts work through a self-contained Joey NMT tutorial, showing that novices perform almost as well as experts in a subsequent code quiz. Joey NMT is available at https://github.com/joeynmt/joeynmt .

arXiv:1907.10761공식 코드 · Papers with Code

Bilingual Lexicon Induction through Unsupervised Machine Translation

A recent research line has obtained strong results on bilingual lexicon induction by aligning independently trained word embeddings in two languages and using the resulting cross-lingual embeddings to induce word translation pairs through nearest neighbor or related retrieval methods. In this paper, we propose an alternative approach to this problem that builds on the recent work on unsupervised machine translation. This way, instead of directly inducing a bilingual lexicon from cross-lingual embeddings, we use them to build a phrase-table, combine it with a language model, and use the resulting machine translation system to generate a synthetic parallel corpus, from which we extract the bilingual lexicon using statistical word alignment techniques. As such, our method can work with any word embedding and cross-lingual mapping technique, and it does not require any additional resource besides the monolingual corpus used to train the embeddings. When evaluated on the exact same cross-lingual embeddings, our proposed method obtains an average improvement of 6 accuracy points over nearest neighbor and 4 points over CSLS retrieval, establishing a new state-of-the-art in the standard MUSE dataset.

arXiv:1907.08710공식 코드 · Papers with Code

Structure-Invariant Testing for Machine Translation

In recent years, machine translation software has increasingly been integrated into our daily lives. People routinely use machine translation for various applications, such as describing symptoms to a foreign doctor and reading political news in a foreign language. However, the complexity and intractability of neural machine translation (NMT) models that power modern machine translation make the robustness of these systems difficult to even assess, much less guarantee. Machine translation systems can return inferior results that lead to misunderstanding, medical misdiagnoses, threats to personal safety, or political conflicts. Despite its apparent importance, validating the robustness of machine translation systems is very difficult and has, therefore, been much under-explored. To tackle this challenge, we introduce structure-invariant testing (SIT), a novel metamorphic testing approach for validating machine translation software. Our key insight is that the translation results of "similar" source sentences should typically exhibit similar sentence structures. Specifically, SIT (1) generates similar source sentences by substituting one word in a given sentence with semantically similar, syntactically equivalent words; (2) represents sentence structure by syntax parse trees (obtained via constituency or dependency parsing); (3) reports sentence pairs whose structures differ quantitatively by more than some threshold. To evaluate SIT, we use it to test Google Translate and Bing Microsoft Translator with 200 source sentences as input, which led to 64 and 70 buggy issues with 69.5\% and 70\% top-1 accuracy, respectively. The translation errors are diverse, including under-translation, over-translation, incorrect modification, word/phrase mistranslation, and unclear logic.

arXiv:1907.09282공식 코드 · Papers with Code

Learning the Relation between Code Features and Code Transforms with Structured Prediction

To effectively guide the exploration of the code transform space for automated code evolution techniques, we present in this paper the first approach for structurally predicting code transforms at the level of AST nodes using conditional random fields (CRFs). Our approach first learns offline a probabilistic model that captures how certain code transforms are applied to certain AST nodes, and then uses the learned model to predict transforms for arbitrary new, unseen code snippets. {Our approach involves a novel representation of both programs and code transforms. Specifically, we introduce the formal framework for defining the so-called AST-level code transforms and we demonstrate how the CRF model can be accordingly designed, learned, and used for prediction}. We instantiate our approach in the context of repair transform prediction for Java programs. Our instantiation contains a set of carefully designed code features, deals with the training data imbalance issue, and comprises transform constraints that are specific to code. We conduct a large-scale experimental evaluation based on a dataset of bug fixing commits from real-world Java projects. The results show that when the popular evaluation metric \emph{top-3} is used, our approach predicts the code transforms with an accuracy varying from 41\% to 53\% depending on the transforms. Our model outperforms two baselines based on history probability and neural machine translation (NMT), suggesting the importance of considering code structure in achieving good prediction accuracy. In addition, a proof-of-concept synthesizer is implemented to concretize some repair transforms to get the final patches. The evaluation of the synthesizer on the Defects4j benchmark confirms the usefulness of the predicted AST-level repair transforms in producing high-quality patches.

arXiv:1907.08610공식 코드 · Papers with Code

Lookahead Optimizer: k steps forward, 1 step back

The vast majority of successful deep neural networks are trained using variants of stochastic gradient descent (SGD) algorithms. Recent attempts to improve SGD can be broadly categorized into two approaches: (1) adaptive learning rate schemes, such as AdaGrad and Adam, and (2) accelerated schemes, such as heavy-ball and Nesterov momentum. In this paper, we propose a new optimization algorithm, Lookahead, that is orthogonal to these previous approaches and iteratively updates two sets of weights. Intuitively, the algorithm chooses a search direction by looking ahead at the sequence of fast weights generated by another optimizer. We show that Lookahead improves the learning stability and lowers the variance of its inner optimizer with negligible computation and memory cost. We empirically demonstrate Lookahead can significantly improve the performance of SGD and Adam, even with their default hyperparameter settings on ImageNet, CIFAR-10/100, neural machine translation, and Penn Treebank.

arXiv:1907.08506공식 코드 · Papers with Code

Language Modelling for Sound Event Detection with Teacher Forcing and Scheduled Sampling

A sound event detection (SED) method typically takes as an input a sequence of audio frames and predicts the activities of sound events in each frame. In real-life recordings, the sound events exhibit some temporal structure: for instance, a "car horn" will likely be followed by a "car passing by". While this temporal structure is widely exploited in sequence prediction tasks (e.g., in machine translation), where language models (LM) are exploited, it is not satisfactorily modeled in SED. In this work we propose a method which allows a recurrent neural network (RNN) to learn an LM for the SED task. The method conditions the input of the RNN with the activities of classes at the previous time step. We evaluate our method using F1 score and error rate (ER) over three different and publicly available datasets; the TUT-SED Synthetic 2016 and the TUT Sound Events 2016 and 2017 datasets. The obtained results show an increase of 9% and 2% at the F1 (higher is better) and a decrease of 7% and 2% at ER (lower is better) for the TUT Sound Events 2016 and 2017 datasets, respectively, when using our method. On the contrary, with our method there is a decrease of 4% at F1 score and an increase of 7% at ER for the TUT-SED Synthetic 2016 dataset.

arXiv:1907.06142공식 코드 · Papers with Code

Tackling Graphical NLP problems with Graph Recurrent Networks

How to properly model graphs is a long-existing and important problem in NLP area, where several popular types of graphs are knowledge graphs, semantic graphs and dependency graphs. Comparing with other data structures, such as sequences and trees, graphs are generally more powerful in representing complex correlations among entities. For example, a knowledge graph stores real-word entities (such as "Barack_Obama" and "U.S.") and their relations (such as "live_in" and "lead_by"). Properly encoding a knowledge graph is beneficial to user applications, such as question answering and knowledge discovery. Modeling graphs is also very challenging, probably because graphs usually contain massive and cyclic relations. Recent years have witnessed the success of deep learning, especially RNN-based models, on many NLP problems. Besides, RNNs and their variations have been extensively studied on several graph problems and showed preliminary successes. Despite the successes that have been achieved, RNN-based models suffer from several major drawbacks on graphs. First, they can only consume sequential data, thus linearization is required to serialize input graphs, resulting in the loss of important structural information. Second, the serialization results are usually very long, so it takes a long time for RNNs to encode them. In this thesis, we propose a novel graph neural network, named graph recurrent network (GRN). We study our GRN model on 4 very different tasks, such as machine reading comprehension, relation extraction and machine translation. Some take undirected graphs without edge labels, while the others have directed ones with edge labels. To consider these important differences, we gradually enhance our GRN model, such as further considering edge labels and adding an RNN decoder. Carefully designed experiments show the effectiveness of GRN on all these tasks.

arXiv:1907.05190공식 코드 · Papers with Code

Self-Regulated Interactive Sequence-to-Sequence Learning

Not all types of supervision signals are created equal: Different types of feedback have different costs and effects on learning. We show how self-regulation strategies that decide when to ask for which kind of feedback from a teacher (or from oneself) can be cast as a learning-to-learn problem leading to improved cost-aware sequence-to-sequence learning. In experiments on interactive neural machine translation, we find that the self-regulator discovers an $\epsilon$-greedy strategy for the optimal cost-quality trade-off by mixing different feedback types including corrections, error markups, and self-supervision. Furthermore, we demonstrate its robustness under domain shift and identify it as a promising alternative to active learning.

arXiv:1907.03748공식 코드 · Papers with Code

Learning Neural Sequence-to-Sequence Models from Weak Feedback with Bipolar Ramp Loss

In many machine learning scenarios, supervision by gold labels is not available and consequently neural models cannot be trained directly by maximum likelihood estimation (MLE). In a weak supervision scenario, metric-augmented objectives can be employed to assign feedback to model outputs, which can be used to extract a supervision signal for training. We present several objectives for two separate weakly supervised tasks, machine translation and semantic parsing. We show that objectives should actively discourage negative outputs in addition to promoting a surrogate gold structure. This notion of bipolarity is naturally present in ramp loss objectives, which we adapt to neural models. We show that bipolar ramp loss objectives outperform other non-bipolar ramp loss objectives and minimum risk training (MRT) on both weakly supervised tasks, as well as on a supervised machine translation task. Additionally, we introduce a novel token-level ramp loss objective, which is able to outperform even the best sequence-level ramp loss on both weakly supervised tasks.

arXiv:1907.01968공식 코드 · Papers with Code

Depth Growing for Neural Machine Translation

While very deep neural networks have shown effectiveness for computer vision and text classification applications, how to increase the network depth of neural machine translation (NMT) models for better translation quality remains a challenging problem. Directly stacking more blocks to the NMT model results in no improvement and even reduces performance. In this work, we propose an effective two-stage approach with three specially designed components to construct deeper NMT models, which result in significant improvements over the strong Transformer baselines on WMT$14$ English$\to$German and English$\to$French translation tasks\footnote{Our code is available at \url{https://github.com/apeterswu/Depth_Growing_NMT}}.

arXiv:1907.00900공식 코드 · Papers with Code

Post-editese: an Exacerbated Translationese

Post-editing (PE) machine translation (MT) is widely used for dissemination because it leads to higher productivity than human translation from scratch (HT). In addition, PE translations are found to be of equal or better quality than HTs. However, most such studies measure quality solely as the number of errors. We conduct a set of computational analyses in which we compare PE against HT on three different datasets that cover five translation directions with measures that address different translation universals and laws of translation: simplification, normalisation and interference. We find out that PEs are simpler and more normalised and have a higher degree of interference from the source language than HTs.

arXiv:1907.00184공식 코드 · Papers with Code

Empirical Evaluation of Sequence-to-Sequence Models for Word Discovery in Low-resource Settings

Since Bahdanau et al. [1] first introduced attention for neural machine translation, most sequence-to-sequence models made use of attention mechanisms [2, 3, 4]. While they produce soft-alignment matrices that could be interpreted as alignment between target and source languages, we lack metrics to quantify their quality, being unclear which approach produces the best alignments. This paper presents an empirical evaluation of 3 main sequence-to-sequence models (CNN, RNN and Transformer-based) for word discovery from unsegmented phoneme sequences. This task consists in aligning word sequences in a source language with phoneme sequences in a target language, inferring from it word segmentation on the target side [5]. Evaluating word segmentation quality can be seen as an extrinsic evaluation of the soft-alignment matrices produced during training. Our experiments in a low-resource scenario on Mboshi and English languages (both aligned to French) show that RNNs surprisingly outperform CNNs and Transformer for this task. Our results are confirmed by an intrinsic evaluation of alignment quality through the use of Average Normalized Entropy (ANE). Lastly, we improve our best word discovery model by using an alignment entropy confidence measure that accumulates ANE over all the occurrences of a given alignment pair in the collection.

arXiv:1906.12284공식 코드 · Papers with Code

Widening the Representation Bottleneck in Neural Machine Translation with Lexical Shortcuts

The transformer is a state-of-the-art neural translation model that uses attention to iteratively refine lexical representations with information drawn from the surrounding context. Lexical features are fed into the first layer and propagated through a deep network of hidden layers. We argue that the need to represent and propagate lexical features in each layer limits the model's capacity for learning and representing other information relevant to the task. To alleviate this bottleneck, we introduce gated shortcut connections between the embedding layer and each subsequent layer within the encoder and decoder. This enables the model to access relevant lexical content dynamically, without expending limited resources on storing it within intermediate states. We show that the proposed modification yields consistent improvements over a baseline transformer on standard WMT translation tasks in 5 translation directions (0.9 BLEU on average) and reduces the amount of lexical information passed along the hidden layers. We furthermore evaluate different ways to integrate lexical connections into the transformer architecture and present ablation experiments exploring the effect of proposed shortcuts on model behavior.

arXiv:1906.11943공식 코드 · Papers with Code

Findings of the First Shared Task on Machine Translation Robustness

We share the findings of the first shared task on improving robustness of Machine Translation (MT). The task provides a testbed representing challenges facing MT models deployed in the real world, and facilitates new approaches to improve models; robustness to noisy input and domain mismatch. We focus on two language pairs (English-French and English-Japanese), and the submitted systems are evaluated on a blind test set consisting of noisy comments on Reddit and professionally sourced translations. As a new task, we received 23 submissions by 11 participating teams from universities, companies, national labs, etc. All submitted systems achieved large improvements over baselines, with the best improvement having +22.33 BLEU. We evaluated submissions by both human judgment and automatic evaluation (BLEU), which shows high correlations (Pearson's r = 0.94 and 0.95). Furthermore, we conducted a qualitative analysis of the submitted systems using compare-mt, which revealed their salient differences in handling challenges in this task. Such analysis provides additional insights when there is occasional disagreement between human judgment and BLEU, e.g. systems better at producing colloquial expressions received higher score from human judgment.

arXiv:1906.10519공식 코드 · Papers with Code

Embedding Projection for Targeted Cross-Lingual Sentiment: Model Comparisons and a Real-World Study

Sentiment analysis benefits from large, hand-annotated resources in order to train and test machine learning models, which are often data hungry. While some languages, e.g., English, have a vast array of these resources, most under-resourced languages do not, especially for fine-grained sentiment tasks, such as aspect-level or targeted sentiment analysis. To improve this situation, we propose a cross-lingual approach to sentiment analysis that is applicable to under-resourced languages and takes into account target-level information. This model incorporates sentiment information into bilingual distributional representations, by jointly optimizing them for semantics and sentiment, showing state-of-the-art performance at sentence-level when combined with machine translation. The adaptation to targeted sentiment analysis on multiple domains shows that our model outperforms other projection-based bilingual embedding methods on binary targeted sentiment tasks. Our analysis on ten languages demonstrates that the amount of unlabeled monolingual data has surprisingly little effect on the sentiment results. As expected, the choice of annotated source language for projection to a target leads to better results for source-target language pairs which are similar. Therefore, our results suggest that more efforts should be spent on the creation of resources for less similar languages to those which are resource-rich already. Finally, a domain mismatch leads to a decreased performance. This suggests resources in any language should ideally cover varieties of domains.