Dominican Republic

Multi-head attention, a collection of several attention mechanisms that independently attend to different parts of the input, is the key ingredient in the Transformer. Recent work has shown, however, that a large proportion of the heads in a Transformer&#39;s multi-head attention mechanism can be safely pruned away without significantly harming the performance of the model; such pruning leads to models that are noticeably smaller and faster in practice. Our work introduces a new head pruning technique that we term differentiable subset pruning. Intuitively, our method learns per-head importance variables and then enforces a user-specified hard constraint on the number of unpruned heads. The importance variables are learned via stochastic gradient descent. We conduct experiments on natural language inference and machine translation; we show that differentiable subset pruning performs comparably or better than previous works while offering precise control of the sparsity level.

EMNLP 2021

Differentiable Subset Pruning of Transformer Heads

EMNLP 2021 is planned to be a hybrid event in Punta Cana, Dominican Republic, with both on-site and fully virtual participation possible. The experience for on-site participants would closely approximate a normal pre-COVID *ACL conference, with 5-6 thematically organized parallel sessions and live Q/A and interactive discussion immediately after the talks. Presentations by virtual participants will be equitably interleaved with those of on-site participants, projected on the auditorium screens as if on-site, and also followed immediately by live Q/A and interactive discussion at a time during reasonable waking hours for the virtual presenter. For all participants, on-site and virtual, who are unable to attend a session due to either time-zone issues or because they are participating in another session live, talk recordings and slides will be available online at a minimum after the live presentation (and in many cases before as well), and questions may be submitted in advance on session-specific discussion boards and answered live in session with the usual visual aids if desired.

<iframe style="width:700px;height:400px" src="https://online.fliphtml5.com/ebtyf/ceby/"  seamless="seamless" scrolling="no" frameborder="0" allowtransparency="true" allowfullscreen="true" ></iframe>

Please Note: The EMNLP registration system is not currently connected to the underline site as we are still in the process of building out EMNLP 2021. You will receive access instructions from underline the week of November 1st. 

Access is given only to EMNLP upon registration, if you have not registered please do so [here](https://2021.emnlp.org/registration).

Registered attendees will receive access the week of November 1st.

EMNLP 2021 is planned to be a hybrid event in Punta Cana, Dominican Republic, with both on-site and fully virtual participation possible.

Multi-head attention, a collection of several attention mechanisms that independently attend to different parts of the input, is the key ingredient in the Transformer. Recent work has shown, however, that a large proportion of the heads in a Transformer's multi-head attention mechanism can be safely pruned away without significantly harming the performance of the model; such pruning leads to models that are noticeably smaller and faster in practice. Our work introduces a new head pruning technique that we term differentiable subset pruning. Intuitively, our method learns per-head importance variables and then enforces a user-specified hard constraint on the number of unpruned heads. The importance variables are learned via stochastic gradient descent. We conduct experiments on natural language inference and machine translation; we show that differentiable subset pruning performs comparably or better than previous works while offering precise control of the sparsity level.

poster

Masked language modeling (MLM), a self-supervised pretraining objective, is widely used in natural language processing for learning text representations. MLM trains a model to predict a random sample of input tokens that have been replaced by a [MASK] placeholder in a multi-class setting over the entire vocabulary. When pretraining, it is common to use alongside MLM other auxiliary objectives on the token or sequence level to improve downstream performance (e.g. next sentence prediction). However, no previous work so far has attempted in examining whether other simpler linguistically intuitive or not objectives can be used standalone as main pretraining objectives. In this paper, we explore five simple pretraining objectives based on token-level classification tasks as replacements of MLM. Empirical results on GLUE and SQuAD show that our proposed methods achieve comparable or better performance to MLM using a BERT-BASE architecture. We further validate our methods using smaller models, showing that pretraining a model with 41% of the BERT-BASE's parameters, BERT-MEDIUM results in only a 1% drop in GLUE scores with our best objective.

Frustratingly Simple Pretraining Alternatives to Masked Language Modeling

Recent studies have shown that deep neural network-based models are vulnerable to intentionally crafted adversarial examples, and various methods have been proposed to defend against adversarial word-substitution attacks for neural NLP models. However, there is a lack of systematic study on comparing different defense approaches under the same attacking setting. In this paper, we seek to fill the gap through comprehensive studies on the behavior of neural text classifiers trained with various defense methods against representative adversarial attacks. In addition, we propose an effective method to further improve the robustness of neural text classifiers against such attacks, and achieved the highest accuracy on both clean and adversarial examples on AGNEWS and IMDB datasets, outperforming existing methods by a significant margin. We hope this study could provide useful clues for future research on text adversarial defense. Codes are available at https://github.com/RockyLzy/TextDefender.

Searching for an Effective Defender: Benchmarking Defense against Adversarial Word Substitution

Difﬁcult samples of the minority class in imbalanced text classiﬁcation are usually hard to be classiﬁed as they are embedded into an overlapping semantic region with the majority class. In this paper, we propose a Mutual Information constrained Semantically Oversampling framework (MISO) that can generate anchor instances to help the backbone network determine the re-embedding position of a non-overlapping representation for each difﬁcult sample. MISO consists of (1) a semantic fusion module that learns entangled semantics among difﬁcult and majority samples with an adaptive multi-head attention mechanism, (2) a mutual information loss that forces our model to learn new representations of entangled semantics in the non-overlapping region of the minority class, and (3) a coupled adversarial encoder-decoder that ﬁne-tunes disentangled semantic representations to remain their correlations with the minority class, and then using these disentangled semantic representations to generate anchor instances for each difﬁcult sample. Experiments on a variety of imbalanced text classiﬁcation tasks demonstrate that anchor instances help classiﬁers achieve signiﬁcant improvements over strong baselines.

Re-embedding Difficult Samples via Mutual Information Constrained Semantically Oversampling for Imbalanced Text Classification

Recent research has investigated quantum NLP, designing algorithms that process natural language in quantum computers, and also quantum-inspired algorithms that improve NLP performance on classical computers. In this survey, we review representative methods at the intersection of NLP and quantum physics in the past ten years, categorizing them according to the use of quantum theory, the linguistic targets that are modeled, and the downstream application. The literature review ends with a discussion on the key factors to the success that has been achieved by existing work, as well as challenges ahead, with the goal of better understanding the promises and further directions.

Natural Language Processing Meets Quantum Physics: A Survey and Categorization

Neural models and symbolic rules such as regular expressions have their respective merits and weaknesses. In this paper, we study the integration of the two approaches for the slot filling task by converting regular expressions into neural networks. Specifically, we first convert regular expressions into a special form of finite-state transducers, then unfold its approximate inference algorithm as a bidirectional recurrent neural model that performs slot filling via sequence labeling. Experimental results show that our model has superior zero-shot and few-shot performance and stays competitive when there are sufficient training data.

Neuralizing Regular Expressions for Slot Filling

Knowledge distillation (KD) is a common knowledge transfer algorithm used for model compression across a variety of deep learning based natural language processing (NLP) solutions. In its regular manifestations, KD requires access to the teacher’s training data for knowledge transfer to the student network. However, privacy concerns, data regulations and proprietary reasons may prevent access to such data. We present, to the best of our knowledge, the first work on Zero-shot Knowledge Distillation for NLP, where the student learns from the much larger teacher without any task specific data. Our solution combines out-of-domain data and adversarial training to learn the teacher’s output distribution. We investigate six tasks from the GLUE benchmark and demonstrate that we can achieve between 75% and 92% of the teacher’s classification score (accuracy or F1) while compressing the model 30 times.

Towards Zero-Shot Knowledge Distillation for Natural Language Processing

Language models trained on billions of tokens have recently led to unprecedented results on many NLP tasks. This success raises the question of whether, in principle, a system can ever "understand" raw text without access to some form of grounding. We formally investigate the abilities of ungrounded systems to acquire meaning. Our analysis focuses on the role of "assertions": textual contexts that provide indirect clues about the underlying semantics. We study whether assertions enable a system to emulate representations preserving semantic relations like equivalence. We find that assertions enable semantic emulation of languages that satisfy a strong notion of semantic transparency. However, for classes of languages where the same expression can take different values in different contexts, we show that emulation can become uncomputable. Finally, we discuss differences between our formal model and natural language, exploring how our results generalize to a modal setting and other semantic relations. Together, our results suggest that assertions in code or language do not provide sufficient signal to fully emulate semantic representations. We formalize ways in which ungrounded language models appear to be fundamentally limited in their ability to "understand".

Provable Limitations of Acquiring Meaning from Ungrounded Form: What will Future Language Models Understand?

The paper reports the results of a translationese study of literary texts based on translated and non-translated Russian. We aim to find out if translations deviate from non-translated literary texts, and if the established differences can be attributed to typological relations between source and target languages. We expect that literary translations from typologically distant languages should exhibit more translationese, and the fingerprints of individual source languages (and their families) are traceable in translations. We explore linguistic properties that distinguish non-translated Russian literature from translations into Russian. Our results show that non-translated fiction is different from translations to the degree that these two language varieties can be automatically classified. As expected, language typology is reflected in translations of literary texts. We identified features that point to the linguistic specificity of Russian non-translated literature and to shining-through effects. Some of the translationese features cut across all language pairs, while others are characteristic of literary translations from languages belonging to specific language families.

Translationese in Russian Literary Texts

The dawn of the digital age led to increasing demands for digital research resources, which shall be quickly processed and handled by computers. Due to the amount of data created by this digitization process, the design of tools that enable the analysis and management of data and metadata has become a relevant topic. In this context, the Multilingual Corpus of Survey Questionnaires (MCSQ) contributes to the creation and distribution of data for the Social Sciences and Humanities (SSH) following FAIR (Findable, Accessible, Interoperable and Reusable) principles, and provides functionalities for end-users that are not acquainted with programming through an easy-to-use interface. By simply applying the desired filters in the graphic interface, users can build linguistic resources for the survey research and translation areas, such as translation memories, thus facilitating data access and usage.

The Multilingual Corpus of Survey Questionnaires Query Interface

This paper presents a data set of German fairy tales, manually annotated with character networks which were obtained with high interrater agreement. The release of this corpus provides an opportunity of training and comparing different algorithms for the extraction of character networks, which so far was barely possible due to heterogeneous interests of previous researchers. We demonstrate the usefulness of our data set by providing baseline experiments for the automatic extraction of character networks, applying a rule-based pipeline as well as a neural approach, and find the neural approach outperforming the rule-approach in most evaluation settings.

Downloads

Next from EMNLP 2021

Frustratingly Simple Pretraining Alternatives to Masked Language Modeling

Similar lecture

Differentiable Subset Pruning of Transformer Heads

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from EMNLP 2021

Frustratingly Simple Pretraining Alternatives to Masked Language Modeling

Similar lecture

Differentiable Subset Pruning of Transformer Heads

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES

Downloads