ACL 2021 Papers

Named entity recognition (NER) is a well-studied task in natural language processing. Traditional NER research only deals with flat entities and ignores nested entities. The span-based methods treat entity recognition as a span classification task. Although these methods have the innate ability to handle nested NER, they suffer from high computational cost, ignorance of boundary information, under-utilization of the spans that partially match with entities, and difficulties in long entity recognition. To tackle these issues, we propose a two-stage entity identifier. First we generate span proposals by filtering and boundary regression on the seed spans to locate the entities, and then label the boundary-adjusted span proposals with the corresponding categories. Our method effectively utilizes the boundary information of entities and partially matched spans during training. Through boundary regression, entities of any length can be covered theoretically, which improves the ability to recognize long entities. In addition, many low-quality seed spans are filtered out in the first stage, which reduces the time complexity of inference. Experiments on nested NER datasets demonstrate that our proposed method outperforms previous state-of-the-art models.

Large-scale models for learning fixed-dimensional cross-lingual sentence representations like LASER (Artetxe and Schwenk, 2019b) lead to significant improvement in performance on downstream tasks. However, further increases and modifications based on such large-scale models are usually impractical due to memory limitations. In this work, we introduce a lightweight dual-transformer architecture with just 2 layers for generating memory-efficient cross-lingual sentence representations. We explore different training tasks and observe that current cross-lingual training tasks leave a lot to be desired for this shallow architecture. To ameliorate this, we propose a novel cross-lingual language model, which combines the existing single-word masked language model with the newly proposed cross-lingual token-level reconstruction task. We further augment the training task by the introduction of two computationally-lite sentence-level contrastive learning tasks to enhance the alignment of cross-lingual sentence representation space, which compensates for the learning bottleneck of the lightweight transformer for generative tasks. Our comparisons with competing models on cross-lingual sentence retrieval and multilingual document classification confirm the effectiveness of the newly proposed training tasks for a shallow model.

Adapter modules were recently introduced as an efficient alternative to fine-tuning in NLP. Adapter tuning consists in freezing pre-trained parameters of a model and injecting lightweight modules between layers, resulting in the addition of only a small number of task-specific trainable parameters. While adapter tuning was investigated for multilingual neural machine translation, this paper proposes a comprehensive analysis of adapters for multilingual speech translation (ST). Starting from different pre-trained models (a multilingual ST trained on parallel data or a multilingual BART (mBART) trained on non parallel multilingual data), we show that adapters can be used to: (a) efficiently specialize ST to specific language pairs with a low extra cost in terms of parameters, and (b) transfer from an automatic speech recognition (ASR) task and an mBART pre-trained model to a multilingual ST task. Experiments show that adapter tuning offer competitive results to full fine-tuning, while being much more parameter-efficient.

Sequence-to-sequence transduction is the core problem in language processing applications as diverse as semantic parsing, machine translation, and instruction following. The neural network models that provide the dominant solution to these problems are brittle, especially in low-resource settings: they fail to generalize correctly or systematically from small datasets. Past work has shown that many failures of systematic generalization arise from neural models’ inability to disentangle lexical phenomena from syntactic ones. To address this, we augment neural decoders with a lexical translation mechanism that generalizes existing copy mechanisms to incorporate learned, decontextualized, token-level translation rules. We describe how to initialize this mechanism using a variety of lexicon learning algorithms, and show that it improves systematic generalization on a diverse set of sequence modeling tasks drawn from cognitive science, formal semantics, and machine translation.

Lexicon information and pre-trained models, such as BERT, have been combined to explore Chinese sequence labeling tasks due to their respective strengths. However, existing methods solely fuse lexicon features via a shallow and random initialized sequence layer and do not integrate them into the bottom layers of BERT. In this paper, we propose Lexicon Enhanced BERT (LEBERT) for Chinese sequence labeling, which integrates external lexicon knowledge into BERT layers directly by a Lexicon Adapter layer. Compared with existing methods, our model facilitates deep lexicon knowledge fusion at the lower layers of BERT. Experiments on ten Chinese datasets of three tasks including Named Entity Recognition, Word Segmentation, and Part-of-Speech Tagging, show that LEBERT achieves state-of-the-art results.

While there is a large amount of research in the field of Lexical Semantic Change Detection, only few approaches go beyond a standard benchmark evaluation of existing models. In this paper, we propose a shift of focus from change detection to change discovery, i.e., discovering novel word senses over time from the full corpus vocabulary. By heavily fine-tuning a type-based and a token-based approach on recently published German data, we demonstrate that both models can successfully be applied to discover new words undergoing meaning change. Furthermore, we provide an almost fully automated framework for both evaluation and discovery.

Transformer-based language models (LMs) pretrained on large text collections implicitly store a wealth of lexical semantic knowledge, but it is non-trivial to extract that knowledge effectively from their parameters. Inspired by prior work on semantic specialization of static word embedding (WE) models, we show that it is possible to expose and enrich lexical knowledge from the LMs, that is, to specialize them to serve as effective and universal “decontextualized” word encoders even when fed input words “in isolation” (i.e., without any context). Their transformation into such word encoders is achieved through a simple and efficient lexical fine-tuning procedure (termed LexFit) based on dual-encoder network structures. Further, we show that LexFit can yield effective word encoders even with limited lexical supervision and, via cross-lingual transfer, in different languages without any readily available external knowledge. Our evaluation over four established, structurally different lexical-level tasks in 8 languages indicates the superiority of LexFit-based WEs over standard static WEs (e.g., fastText) and WEs from vanilla LMs. Other extensive experiments and ablation studies further profile the LexFit framework, and indicate best practices and performance variations across LexFit variants, languages, and lexical tasks, also directly questioning the usefulness of traditional WE models in the era of large neural models.

A common issue in real-world applications of named entity recognition and classification (NERC) is the absence of annotated data for the target entity classes during training. Zero-shot learning approaches address this issue by learning models from classes with training data that can predict classes without it. This paper presents the first approach for zero-shot NERC, introducing novel architectures that leverage the fact that textual descriptions for many entity classes occur naturally. We address the zero-shot NERC specific challenge that the not-an-entity class is not well defined as different entity classes are considered in training and testing. For evaluation, we adapt two datasets, OntoNotes and MedMentions, emulating the difficulty of real-world zero-shot learning by testing models on the rarest entity classes. Our proposed approach outperforms baselines adapted from machine reading comprehension and zero-shot text classification. Furthermore, we assess the effect of different class descriptions for this task.

Despite transformers’ impressive accuracy, their computational cost is often prohibitive to use with limited computational resources. Most previous approaches to improve inference efficiency require a separate model for each possible computational budget. In this paper, we extend PoWER-BERT (Goyal et al., 2020) and propose Length-Adaptive Transformer that can be used for various inference scenarios after one-shot training. We train a transformer with LengthDrop, a structural variant of dropout, which stochastically determines a sequence length at each layer. We then conduct a multi-objective evolutionary search to find a length configuration that maximizes the accuracy and minimizes the efficiency metric under any given computational budget. Additionally, we significantly extend the applicability of PoWER-BERT beyond sequence-level classification into token-level classification with Drop-and-Restore process that drops word-vectors temporarily in intermediate layers and restores at the last layer if necessary. We empirically verify the utility of the proposed approach by demonstrating the superior accuracy-efficiency trade-off under various setups, including span-based question answering and text classification. Code is available at https://github.com/clovaai/lengthadaptive-transformer.

Pre-trained language models like BERT are performant in a wide range of natural language tasks. However, they are resource exhaustive and computationally expensive for industrial scenarios. Thus, early exits are adopted at each layer of BERT to perform adaptive computation by predicting easier samples with the first few layers to speed up the inference. In this work, to improve efficiency without performance drop, we propose a novel training scheme called Learned Early Exit for BERT (LeeBERT). First, we ask each exit to learn from each other, rather than learning only from the last layer. Second, the weights of different loss terms are learned, thus balancing off different objectives. We formulate the optimization of LeeBERT as a bi-level optimization problem, and we propose a novel cross-level optimization (CLO) algorithm to improve the optimization results. Experiments on the GLUE benchmark show that our proposed methods improve the performance of the state-of-the-art (SOTA) early exit methods for pre-trained models.

In real scenarios, a multilingual model trained to solve NLP tasks on a set of languages can be required to support new languages over time. Unfortunately, the straightforward retraining on a dataset containing annotated examples for all the languages is both expensive and time-consuming, especially when the number of target languages grows. Moreover, the original annotated material may no longer be available due to storage or business constraints. Re-training only with the new language data will inevitably result in Catastrophic Forgetting of previously acquired knowledge. We propose a Continual Learning strategy that updates a model to support new languages over time, while maintaining consistent results on previously learned languages. We define a Teacher-Student framework where the existing model “teaches” to a student model its knowledge about the languages it supports, while the student is also trained on a new language. We report an experimental evaluation in several tasks including Sentence Classification, Relational Learning and Sequence Labeling.

QA models based on pretrained language models have achieved remarkable performance on various benchmark datasets. However, QA models do not generalize well to unseen data that falls outside the training distribution, due to distributional shifts. Data augmentation (DA) techniques which drop/replace words have shown to be effective in regularizing the model from overfitting to the training data. Yet, they may adversely affect the QA tasks since they incur semantic changes that may lead to wrong answers for the QA task. To tackle this problem, we propose a simple yet effective DA method based on a stochastic noise generator, which learns to perturb the word embedding of the input questions and context without changing their semantics. We validate the performance of the QA models trained with our word embedding perturbation on a single source dataset, on five different target domains. The results show that our method significantly outperforms the baseline DA methods. Notably, the model trained with ours outperforms the model trained with more than 240K artificially generated QA pairs.

Pre-trained text-to-text transformers such as BART have achieved impressive performance across a range of NLP tasks. Recent study further shows that they can learn to generalize to novel tasks, by including task descriptions as part of the source sequence and training the model with (source, target) examples. At test time, these fine-tuned models can make inferences on new tasks using the new task descriptions as part of the input. However, this approach has potential limitations, as the model learns to solve individual (source, target) examples (i.e., at the instance level), instead of learning to solve tasks by taking all examples within a task as a whole (i.e., at the task level). To this end, we introduce Hypter, a framework that improves text-to-text transformer’s generalization ability to unseen tasks by training a hypernetwork to generate task-specific, light-weight adapters from task descriptions. Experiments on ZEST dataset and a synthetic SQuAD dataset demonstrate that Hypter improves upon fine-tuning baselines. Notably, when using BART-Large as the main network, Hypter brings 11.3% comparative improvement on ZEST dataset.

The importance of explaining the outcome of a machine learning model, especially a black-box model, is widely acknowledged. Recent approaches explain an outcome by identifying the contributions of input features to this outcome. In environments involving large black-box models or complex inputs, this leads to computationally demanding algorithms. Further, these algorithms often suffer from low stability, with explanations varying significantly across similar examples. In this paper, we propose a Learning to Explain (L2E) approach that learns the behaviour of an underlying explanation algorithm simultaneously from all training examples. Once the explanation algorithm is distilled into an explainer network, it can be used to explain new instances. Our experiments on three classification tasks, which compare our approach to six explanation algorithms, show that L2E is between 5 and 7.5×10ˆ4 times faster than these algorithms, while generating more stable explanations, and having comparable faithfulness to the black-box model.

Conversational Question Simplification (CQS) aims to simplify self-contained questions into conversational ones by incorporating some conversational characteristics, e.g., anaphora and ellipsis. Existing maximum likelihood estimation based methods often get trapped in easily learned tokens as all tokens are treated equally during training. In this work, we introduce a Reinforcement Iterative Sequence Editing (RISE) framework that optimizes the minimum Levenshtein distance through explicit editing actions. RISE is able to pay attention to tokens that are related to conversational characteristics. To train RISE, we devise an Iterative Reinforce Training (IRT) algorithm with a Dynamic Programming based Sampling (DPS) process to improve exploration. Experimental results on two benchmark datasets show that RISE significantly outperforms state-of-the-art methods and generalizes well on unseen data.

We present a human-and-model-in-the-loop process for dynamically generating datasets and training better performing and more robust hate detection models. We provide a new dataset of 40,000 entries, generated and labelled by trained annotators over four rounds of dynamic data creation. It includes 15,000 challenging perturbations and each hateful entry has fine-grained labels for the type and target of hate. Hateful entries make up 54% of the dataset, which is substantially higher than comparable datasets. We show that model performance is substantially improved using this approach. Models trained on later rounds of data collection perform better on test sets and are harder for annotators to trick. They also have better performance on HateCheck, a suite of functional tests for online hate detection. We provide the code, dataset and annotation guidelines for other researchers to use.

In this paper, we propose Inverse Adversarial Training (IAT) algorithm for training neural dialogue systems to avoid generic responses and model dialogue history better. In contrast to standard adversarial training algorithms, IAT encourages the model to be sensitive to the perturbation in the dialogue history and therefore learning from perturbations. By giving higher rewards for responses whose output probability reduces more significantly when dialogue history is perturbed, the model is encouraged to generate more diverse and consistent responses. By penalizing the model when generating the same response given perturbed dialogue history, the model is forced to better capture dialogue history and generate more informative responses. Experimental results on two benchmark datasets show that our approach can better model dialogue history and generate more diverse and consistent responses. In addition, we point out a problem of the widely used maximum mutual information (MMI) based methods for improving the diversity of dialogue response generation models and demonstrate it empirically.

Few-shot Named Entity Recognition (NER) exploits only a handful of annotations to iden- tify and classify named entity mentions. Pro- totypical network shows superior performance on few-shot NER. However, existing prototyp- ical methods fail to differentiate rich seman- tics in other-class words, which will aggravate overfitting under few shot scenario. To address the issue, we propose a novel model, Mining Undefined Classes from Other-class (MUCO), that can automatically induce different unde- fined classes from the other class to improve few-shot NER. With these extra-labeled unde- fined classes, our method will improve the dis- criminative ability of NER classifier and en- hance the understanding of predefined classes with stand-by semantic knowledge. Experi- mental results demonstrate that our model out- performs five state-of-the-art models in both 1- shot and 5-shots settings on four NER bench- marks. We will release the code upon accep- tance. The source code is released on https: //github.com/shuaiwa16/OtherClassNER.git.

Deep learning models for automatic readability assessment generally discard linguistic features traditionally used in machine learning models for the task. We propose to incorporate linguistic features into neural network models by learning syntactic dense embeddings based on linguistic features. To cope with the relationships between the features, we form a correlation graph among features and use it to learn their embeddings so that similar features will be represented by similar embeddings. Experiments with six data sets of two proficiency levels demonstrate that our proposed methodology can complement BERT-only model to achieve significantly better performances for automatic readability assessment.