Distilling knowledge from a well-trained cumbersome network to a small one has recently become a new research topic, as lightweight neural networks with high performance are particularly in need in various resource-restricted systems. This paper addresses the problem of distilling word embeddings for NLP tasks. We propose an encoding approach to distill task-specific knowledge from a set of high-dimensional embeddings, which can reduce model complexity by a large margin as well as retain high accuracy, showing a good compromise between efficiency and performance. Experiments in two tasks reveal the phenomenon that distilling knowledge from cumbersome embeddings is better than directly training neural networks with small embeddings.
Pre-trained language models have been found to capture a surprisingly rich amount of lexical knowledge, ranging from commonsense properties of everyday concepts to detailed factual knowledge about named entities. Among others, this makes it possible to distill high-quality word vectors from pre-trained language models. However, it is currently unclear to what extent it is possible to distill relation embeddings, i.e. vectors that characterize the relationship between two words. Such relation embeddings are appealing because they can, in principle, encode relational knowledge in a more fine-gra…
Pretrained (language) embeddings are versatile, task-agnostic feature representations of entities, like words, that are central to many machine learning applications. These representations can be enriched through retrofitting, a class of methods that incorporate task-specific domain knowledge encoded as a graph over a subset of these entities. However, existing retrofitting algorithms face two limitations: they overfit the observed graph by failing to represent relationships with missing entities; and they underfit the observed graph by only learning embeddings in Euclidean manifolds, which ca…
A computationally expensive and memory intensive neural network lies behind the recent success of language representation learning. Knowledge distillation, a major technique for deploying such a vast language model in resource-scarce environments, transfers the knowledge on individual word representations learned without restrictions. In this paper, inspired by the recent observations that language representations are relatively positioned and have more semantic knowledge as a whole, we present a new knowledge distillation objective for language representation learning that transfers the conte…
Google Cloud AI（谷歌云人工智能）；McGill University（麦吉尔大学）；Google Research（谷歌研究院）
Natural reading orders of words are crucial for information extraction from form-like documents. Despite recent advances in Graph Convolutional Networks (GCNs) on modeling spatial layout patterns of documents, they have limited ability to capture reading orders of given word-level node representations in a graph. We propose Reading Order Equivariant Positional Encoding (ROPE), a new positional encoding technique designed to apprehend the sequential presentation of words in documents. ROPE generates unique reading order codes for neighboring words relative to the target word given a word-level …
Distilling Knowledge from Pre-trained Language Models via Text Smoothing
Baidu Inc.（百度公司）
本文研究压缩预训练的语言模型，如BERT（Devlin et al.，2019），通过师生知识蒸馏。以前的工作通常迫使学生模型严格模仿教师BERT预测的平滑标签。作为替代方案，我们提出了一种新的BERT蒸馏方法，即，要求教师生成平滑的词ID而不是标签，用于在知识蒸馏中教授学生模型。我们称这种方法为文本平滑。实际上，我们使用BERT中的掩蔽语言模型（MLM）的softmax预测来生成给定文本的单词分布，并使用预测的软单词id来平滑这些输入文本。我们假设平滑后的标签和文本都可以隐式地扩充输入语料库，而文本平滑由于可以在一个神经网络前向步骤中生成更多的实例，因此直观上更有效，在GLUE和SQuAD上的实验结果表明，与现有的BERT提取方法相比，我们的解决方案可以获得有竞争力的结果。
The adversarial attacks against deep neural networks on computer vision tasks have spawned many new technologies that help protect models from avoiding false predictions. Recently, word-level adversarial attacks on deep models of Natural Language Processing (NLP) tasks have also demonstrated strong power, e.g., fooling a sentiment classification neural network to make wrong decisions. Unfortunately, few previous literatures have discussed the defense of such word-level synonym substitution based attacks since they are hard to be perceived and detected. In this paper, we shed light on this prob…
Which Encoding is the Best for Text Classification in Chinese, English, Japanese and Korean?
Courant Institute of Mathematical Sciences, New York University（纽约大学库朗数学科学研究所）；Center for Data Science, New York University（纽约大学数据科学中心）；Facebook AI Research, Facebook Inc.（Facebook 公司人工智能研究院）
With the advent of deep learning, a huge number of text-to-speech (TTS) models which produce human-like speech have emerged. Recently, by introducing syntactic and semantic information w.r.t the input text, various approaches have been proposed to enrich the naturalness and expressiveness of TTS models. Although these strategies showed impressive results, they still have some limitations in utilizing language information. First, most approaches only use graph networks to utilize syntactic and semantic information without considering linguistic features. Second, most previous works do not expli…
Graduate School of Engineering（工程研究生院）；The University of Tokyo（东京大学）
In this paper, we tackle the task of definition modeling, where the goal is to learn to generate definitions of words and phrases. Existing approaches for this task are discriminative, combining distributional and lexical semantics in an implicit rather than direct way. To tackle this issue we propose a generative model for the task, introducing a continuous latent variable to explicitly model the underlying relationship between a phrase used within a context and its definition. We rely on variational inference for estimation and leverage contextualized word embeddings for improved performance…
Spoken Language Understanding (SLU) converts hypotheses from automatic speech recognizer (ASR) into structured semantic representations. ASR recognition errors can severely degenerate the performance of the subsequent SLU module. To address this issue, word confusion networks (WCNs) have been used to encode the input for SLU, which contain richer information than 1-best or n-best hypotheses list. To further eliminate ambiguity, the last system act of dialogue context is also utilized as additional input. In this paper, a novel BERT based SLU model (WCN-BERT SLU) is proposed to encode WCNs and …
A Multilingual Encoding Method for Text Classification and Dialect Identification Using Convolutional Neural Network
Arab Academy for Science, Technology, and Maritime Transport（阿拉伯科学、技术与海运学院）；College of Computing & Information Technology（计算与信息技术学院）；Columbus State University（哥伦布州立大学）；TSYS School of Computer Science（TSYS计算机学院）
Peking University（北京大学）；Huazhong University of Science and Technology（华中科技大学）；Microsoft Research（微软研究院）
Transformer has demonstrated its great power to learn contextual word representations for multiple languages in a single model. To process multilingual sentences in the model, a learnable vector is usually assigned to each language, which is called "language embedding". The language embedding can be either added to the word embedding or attached at the beginning of the sentence. It serves as a language-specific signal for the Transformer to capture contextual representations across languages. In this paper, we revisit the use of language embedding and identify several problems in the existing …
University of Amsterdam（阿姆斯特丹大学）；Informatics Institute（信息学研究所）
Neural Machine Translation (NMT) is an open vocabulary problem. As a result, dealing with the words not occurring during training (a.k.a. out-of-vocabulary (OOV) words) have long been a fundamental challenge for NMT systems. The predominant method to tackle this problem is Byte Pair Encoding (BPE) which splits words, including OOV words, into sub-word segments. BPE has achieved impressive results for a wide range of translation tasks in terms of automatic evaluation metrics. While it is often assumed that by using BPE, NMT systems are capable of handling OOV words, the effectiveness of BPE in …
École Normale Supérieure-PSL（巴黎高等师范学院-巴黎文理研究大学）；CNRS（法国国家科学研究中心）；U. Sorbonne Nouvelle（新索邦大学）；University of Pisa（比萨大学）
Word order, an essential property of natural languages, is injected in Transformer-based neural language models using position encoding. However, recent experiments have shown that explicit position encoding is not always useful, since some models without such feature managed to achieve state-of-the art performance on some tasks. To understand better this phenomenon, we examine the effect of removing position encodings on the pre-training objective itself (i.e., masked language modelling), to test whether models can reconstruct position information from co-occurrences alone. We do so by contro…
In this work, we investigate the positional encoding methods used in language pre-training (e.g., BERT) and identify several problems in the existing formulations. First, we show that in the absolute positional encoding, the addition operation applied on positional embeddings and word embeddings brings mixed correlations between the two heterogeneous information resources. It may bring unnecessary randomness in the attention and further limit the expressiveness of the model. Second, we question whether treating the position of the symbol \texttt{[CLS]} the same as other words is a reasonable d…
The vector representation of words, known as word embeddings, has opened a new research approach in linguistic studies. These representations can capture different types of information about words. The grammatical gender of nouns is a typical classification of nouns based on their formal and semantic properties. The study of grammatical gender based on word embeddings can give insight into discussions on how grammatical genders are determined. In this study, we compare different sets of word embeddings according to the accuracy of a neural classifier determining the grammatical gender of nouns…
Incorporating biasing words obtained as contextual knowledge is critical for many automatic speech recognition (ASR) applications. This paper proposes the use of graph neural network (GNN) encodings in a tree-constrained pointer generator (TCPGen) component for end-to-end contextual ASR. By encoding the biasing words in the prefix-tree with a tree-based GNN, lookahead for future wordpieces in end-to-end ASR decoding is achieved at each tree node by incorporating information about all wordpieces on the tree branches rooted from it, which allows a more accurate prediction of the generation proba…
University of Oslo（奥斯陆大学）；University of Copenhagen（哥本哈根大学）
In order to preserve word-order information in a non-autoregressive setting, transformer architectures tend to include positional knowledge, by (for instance) adding positional encodings to token embeddings. Several modifications have been proposed over the sinusoidal positional encodings used in the original transformer architecture; these include, for instance, separating position encodings and token embeddings, or directly modifying attention weights based on the distance between word pairs. We first show that surprisingly, while these modifications tend to improve monolingual language mode…
IIIT Sri City（印度信息技术学院斯里城）；UCLA（加州大学洛杉矶分校）；Wipro AI Labs（威普罗人工智能实验室）；AI Institute, University of South Carolina（南卡罗来纳大学人工智能研究所）
NLP applications for code-mixed (CM) or mix-lingual text have gained a significant momentum recently, the main reason being the prevalence of language mixing in social media communications in multi-lingual societies like India, Mexico, Europe, parts of USA etc. Word embeddings are basic build-ing blocks of any NLP system today, yet, word embedding for CM languages is an unexplored territory. The major bottleneck for CM word embeddings is switching points, where the language switches. These locations lack in contextually and statistical systems fail to model this phenomena due to high variance …
University of Padua（帕多瓦大学）；Tianjin University（天津大学）；University of Copenhagen（哥本哈根大学）
Sequential word order is important when processing text. Currently, neural networks (NNs) address this by modeling word position using position embeddings. The problem is that position embeddings capture the position of individual words, but not the ordered relationship (e.g., adjacency or precedence) between individual word positions. We present a novel and principled solution for modeling both the global absolute positions of words and their order relationships. Our solution generalizes word embeddings, previously defined as independent vectors, to continuous word functions over a variable (…
We study the problem of integrating syntactic information from constituency trees into a neural model in Frame-semantic parsing sub-tasks, namely Target Identification (TI), FrameIdentification (FI), and Semantic Role Labeling (SRL). We use a Graph Convolutional Network to learn specific representations of constituents, such that each constituent is profiled as the production grammar rule it corresponds to. We leverage these representations to build syntactic features for each word in a sentence, computed as the sum of all the constituents on the path between a word and a task-specific node in…
Universidad Distrital Francisco José de Caldas（弗朗西斯科·何塞·德·卡尔达斯地区大学）
One of the stratagems used to deceive spam filters is to substitute vocables with synonyms or similar words that turn the message unrecognisable by the detection algorithms. In this paper we investigate whether the recent development of language models sensitive to the semantics and context of words, such as Google's BERT, may be useful to overcome this adversarial attack (called "Mad-lib" as per the word substitution game). Using a dataset of 5572 SMS spam messages, we first established a baseline of detection performance using widely known document representation models (BoW and TFIDF) and t…
Recurrent models have been dominating the field of neural machine translation (NMT) for the past few years. Transformers \citep{vaswani2017attention}, have radically changed it by proposing a novel architecture that relies on a feed-forward backbone and self-attention mechanism. Although Transformers are powerful, they could fail to properly encode sequential/positional information due to their non-recurrent nature. To solve this problem, position embeddings are defined exclusively for each time step to enrich word information. However, such embeddings are fixed after training regardless of th…
Distilling Rule-based Knowledge into Large Language Models
Gaoling School of Artificial Intelligence, Renmin University of China（中国人民大学高瓴人工智能学院）；Pattern Recognition Center, WeChat AI, Tencent Inc.（腾讯微信AI模式识别中心）
