Transformers have supplanted recurrent models in a large number of NLP tasks. However, the differences in their abilities to model different syntactic properties remain largely unknown. Past works suggest that LSTMs generalize very well on regular languages and have close connections with counter languages. In this work, we systematically study the ability of Transformers to model such languages as well as the role of its individual components in doing so. We first provide a construction of Transformers for a subclass of counter languages, including well-studied languages such as n-ary Boolean Expressions, Dyck-1, and its generalizations. In experiments, we find that Transformers do well on this subclass, and their learned mechanism strongly correlates with our construction. Perhaps surprisingly, in contrast to LSTMs, Transformers do well only on a subset of regular languages with degrading performance as we make languages more complex according to a well-known measure of complexity. Our analysis also provides insights on the role of self-attention mechanism in modeling certain behaviors and the influence of positional encoding schemes on the learning and generalization abilities of the model.
This paper analyzes three formal models of Transformer encoders that differ in the form of their self-attention mechanism: unique hard attention (UHAT); generalized unique hard attention (GUHAT), which generalizes UHAT; and averaging hard attention (AHAT). We show that UHAT and GUHAT Transformers, viewed as string acceptors, can only recognize formal languages in the complexity class AC$^0$, the class of languages recognizable by families of Boolean circuits of constant depth and polynomial size. This upper bound subsumes Hahn's (2020) results that GUHAT cannot recognize the DYCK languages or …
Logical Languages Accepted by Transformer Encoders with Hard Attention
Institute for Mathematical and Computational Engineering, Universidad Católica de Chile（智利天主教大学数学与计算工程研究所）；IMFD Chile（智利数学与计算工程研究所）；CENIA（智利国家人工智能中心）；TU Kaiserslautern（凯泽斯劳滕工业大学）；Max Planck Institute for Software Systems（马克斯·普朗克软件系统研究所）；Steklov Mathematical Institute of Russian Academy of Sciences（俄罗斯科学院斯捷克洛夫数学研究所）
This document aims to be a self-contained, mathematically precise overview of transformer architectures and algorithms (*not* results). It covers what transformers are, how they are trained, what they are used for, their key architectural components, and a preview of the most prominent models. The reader is assumed to be familiar with basic ML terminology and simpler neural network architectures such as MLPs.
As a fundamental problem in ubiquitous computing and machine learning, sensor-based human activity recognition (HAR) has drawn extensive attention and made great progress in recent years. HAR aims to recognize human activities based on the availability of rich time-series data collected from multi-modal sensors such as accelerometers and gyroscopes. However, recent deep learning methods are focusing on one view of the data, i.e., the temporal view, while shallow methods tend to utilize the hand-craft features for recognition, e.g., the statistics view. In this paper, to extract a better featur…
The automatic detection of humor poses a grand challenge for natural language processing. Transformer-based systems have recently achieved remarkable results on this task, but they usually (1)~were evaluated in setups where serious vs humorous texts came from entirely different sources, and (2)~focused on benchmarking performance without providing insights into how the models work. We make progress in both respects by training and analyzing transformer-based humor recognition models on a recently introduced dataset consisting of minimal pairs of aligned sentences, one serious, the other humoro…
Named Entity Recognition is an information extraction task that serves as a preprocessing step for other natural language processing tasks, such as machine translation, information retrieval, and question answering. Named entity recognition enables the identification of proper names as well as temporal and numeric expressions in an open domain text. For Semitic languages such as Arabic, Amharic, and Hebrew, the named entity recognition task is more challenging due to the heavily inflected structure of these languages. In this paper, we present an Amharic named entity recognition system based o…
Probability Distributions Computed by Autoregressive Transformers
University of Notre Dame（内布拉斯加大学达灵顿分校）；ETH Zürich（苏黎世联邦理工学院）；Max-Planck Institute for Software Systems（马克斯·普朗克软件系统研究所）；University of Kaiserslautern-Landau（凯撒斯劳滕-劳恩堡大学）；San José State University（圣何塞州立大学）
Center of Language and Speech Processing（语言与语音处理中心）；Human Language Technology Center of Excellence, Johns Hopkins University（约翰斯·霍普金斯大学人类语言技术卓越中心）
Dialog acts can be interpreted as the atomic units of a conversation, more fine-grained than utterances, characterized by a specific communicative function. The ability to structure a conversational transcript as a sequence of dialog acts -- dialog act recognition, including the segmentation -- is critical for understanding dialog. We apply two pre-trained transformer models, XLNet and Longformer, to this task in English and achieve strong results on Switchboard Dialog Act and Meeting Recorder Dialog Act corpora with dialog act segmentation error rates (DSER) of 8.4% and 14.2%. To understand t…
Although Bidirectional Encoder Representations from Transformers (BERT) have achieved tremendous success in many natural language processing (NLP) tasks, it remains a black box. A variety of previous works have tried to lift the veil of BERT and understand each layer's functionality. In this paper, we found that surprisingly the output layer of BERT can reconstruct the input sentence by directly taking each layer of BERT as input, even though the output layer has never seen the input other than the final hidden layer. This fact remains true across a wide variety of BERT-based models, even when…
While recurrent models have been effective in NLP tasks, their performance on context-free languages (CFLs) has been found to be quite weak. Given that CFLs are believed to capture important phenomena such as hierarchical structure in natural languages, this discrepancy in performance calls for an explanation. We study the performance of recurrent models on Dyck-n languages, a particularly important and well-studied class of CFLs. We find that while recurrent models generalize nearly perfectly if the lengths of the training and test strings are from the same range, they perform poorly if the t…
Allen Institute for AI（艾伦人工智能研究所）；New York University（纽约大学）；University of Washington（华盛顿大学）
Transformers have become a standard neural network architecture for many NLP problems, motivating theoretical analysis of their power in terms of formal languages. Recent work has shown that transformers with hard attention are quite limited in power (Hahn, 2020), as they can be simulated by constant-depth AND/OR circuits (Hao et al. 2021). However, hard attention is a strong assumption, which may complicate the relevance of these results in practice. In this work, we analyze the circuit complexity of transformers with saturated attention: a generalization of hard attention that more closely c…
Czech is a very specific language due to its large differences between the formal and the colloquial form of speech. While the formal (written) form is used mainly in official documents, literature, and public speeches, the colloquial (spoken) form is used widely among people in casual speeches. This gap introduces serious problems for ASR systems, especially when training or evaluating ASR models on datasets containing a lot of colloquial speech, such as the MALACH project. In this paper, we are addressing this problem in the light of a new paradigm in end-to-end ASR systems -- recently intro…
This report is a survey of the relationships between various state-of-the-art neural network architectures and formal languages as, for example, structured by the Chomsky Language Hierarchy. Of particular interest are the abilities of a neural architecture to represent, recognize and generate words from a specific language by learning from samples of the language.
Although transformers are remarkably effective for many tasks, there are some surprisingly easy-looking regular languages that they struggle with. Hahn shows that for languages where acceptance depends on a single input symbol, a transformer's classification decisions become less and less confident (that is, with cross-entropy approaching 1 bit per string) as input strings get longer and longer. We examine this limitation using two languages: PARITY, the language of bit strings with an odd number of 1s, and FIRST, the language of bit strings starting with a 1. We demonstrate three ways of over…
Lattices are compact representations that encode multiple hypotheses, such as speech recognition results or different word segmentations. It is shown that encoding lattices as opposed to 1-best results generated by automatic speech recognizer (ASR) boosts the performance of spoken language understanding (SLU). Recently, pretrained language models with the transformer architecture have achieved the state-of-the-art results on natural language understanding, but their ability of encoding lattices has not been explored. Therefore, this paper aims at adapting pretrained transformers to lattice inp…
Detecting Text Formality: A Study of Text Classification Approaches
Technical University of Munich（慕尼黑工业大学）；Centro Singular de Investigación en Tecnoloxías Intelixentes (CiTIUS)（智能技术单一研究中心）；Universidade de Santiago de Compostela（圣地亚哥德孔波斯特拉大学）；Skolkovo Institute of Science and Technology（斯科尔科沃科技学院）；Artificial Intelligence Research Institute（人工智能研究所）
We develop a chatbot using Deep Bidirectional Transformer models (BERT) to handle client questions in financial investment customer service. The bot can recognize 381 intents, and decides when to say "I don't know" and escalates irrelevant/uncertain questions to human operators. Our main novel contribution is the discussion about uncertainty measure for BERT, where three different approaches are systematically compared on real problems. We investigated two uncertainty metrics, information entropy and variance of dropout sampling in BERT, followed by mixed-integer programming to optimize decisi…
Quasi-symbolic Semantic Geometry over Transformer-based Variational AutoEncoder
Department of Computer Science, University of Manchester（曼彻斯特大学计算机科学系）；Idiap Research Institute（Idiap研究机构）；National Biomarker Centre, CRUK-MI, Univ. of Manchester（国家生物标记中心，CRUK-MI，曼彻斯特大学）
Beginning with McCarthy's Advice Taker (1959), AI has pursued the goal of providing a system with explicit, general knowledge and having the system reason over that knowledge. However, expressing the knowledge in a formal (logical or probabilistic) representation has been a major obstacle to this research. This paper investigates a modern approach to this problem where the facts and rules are provided as natural language sentences, thus bypassing a formal representation. We train transformers to reason (or emulate reasoning) over these sentences using synthetically generated data. Our models, …
Google Research（谷歌研究院）；Seoul National University（首尔国立大学）
Multimodal learning can benefit from the representation power of pretrained Large Language Models (LLMs). However, state-of-the-art transformer based LLMs often ignore negations in natural language and there is no existing benchmark to quantitatively evaluate whether multimodal transformers inherit this weakness. In this study, we present a new multimodal question answering (QA) benchmark adapted from labeled music videos in AudioSet (Gemmeke et al., 2017) with the goal of systematically evaluating if multimodal transformers can perform complex reasoning to recognize new concepts as negation o…
