United Arab Emirates

Recent work on reducing bias in NLP models usually focuses on protecting or isolating information related to a sensitive attribute (like gender or race). However, when sensitive information is semantically entangled with the task information of the input, e.g., gender information is predictive for a profession, a fair trade-off between task performance and bias mitigation is difficult to achieve. Existing approaches perform this trade-off by eliminating bias information from the latent space, lacking control over how much bias is necessarily required to be removed. We argue that a favorable debiasing method should use sensitive information ‘fairly&#39;, rather than blindly eliminating it (Caliskan et al., 2017; Sun et al., 2019; Bogen et al., 2020). In this work, we provide a novel debiasing algorithm by adjusting
the predictive model&#39;s belief to (1) ignore the sensitive information if it is not useful for the task; (2) use sensitive information minimally as necessary for the prediction (while also incurring a penalty). Experimental results on two text classification tasks (influenced by gender) and an open-ended generation task (influenced by race) indicate that our model achieves a desirable trade-off between debiasing and task performance along with producing debiased rationales as evidence.

EMNLP 2022

Controlling Bias Exposure for Fair Interpretable Predictions

explainability

interpretability

controllability

fairness

bias mitigation

**Welcome to EMNLP 2023!**

On behalf of the EMNLP 2023 Organizing Committee, I extend a warm and heartfelt welcome to all of you to the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP). It is with immense pleasure and excitement that we gather here in Singapore, a vibrant hub of innovation and technological advancement.

The conference program is packed with insightful presentations, thought-provoking workshops, and engaging networking opportunities. In addition to the technical sessions, we have also planned several social events that will provide you with the opportunity to connect with your colleagues.

The conference is held in person at the Resorts World Convention Centre in Singapore, and available on-line with the help of Underline.

For those in Singapore, I hope that you will find time to explore the exciting Sentosa Island, Gardens by the Bay, Singapore Botanic Gardens and the many other attractions unique to Singapore.

Of course, we are grateful to our sponsors and partners for their generous support of EMNLP 2023, whose contributions make it possible for us to host this world-class event.

Yuji Matsumoto (RIKEN AIP) 
EMNLP 2023 General Chair

To access the **EMNLP 2023** event page on Underline, you need to register for the Conference. 
Please follow **[this link](https://2023.emnlp.org/registration/)** for more details.

EMNLP 2023

EMNLP 2023 took place in Singapore from Dec 6th to Dec 10th, 2023.

Recent work on reducing bias in NLP models usually focuses on protecting or isolating information related to a sensitive attribute (like gender or race). However, when sensitive information is semantically entangled with the task information of the input, e.g., gender information is predictive for a profession, a fair trade-off between task performance and bias mitigation is difficult to achieve. Existing approaches perform this trade-off by eliminating bias information from the latent space, lacking control over how much bias is necessarily required to be removed. We argue that a favorable debiasing method should use sensitive information ‘fairly', rather than blindly eliminating it (Caliskan et al., 2017; Sun et al., 2019; Bogen et al., 2020). In this work, we provide a novel debiasing algorithm by adjusting
the predictive model's belief to (1) ignore the sensitive information if it is not useful for the task; (2) use sensitive information minimally as necessary for the prediction (while also incurring a penalty). Experimental results on two text classification tasks (influenced by gender) and an open-ended generation task (influenced by race) indicate that our model achieves a desirable trade-off between debiasing and task performance along with producing debiased rationales as evidence.

findings / work in progress

## Welcome to EMNLP 2022!
I am delighted to welcome you to EMNLP 2022! I believe this conference has been complicated beyond any precedent. Over the past year, it’s been thrilling to see the organization team approach each new puzzle with creativity and enthusiasm. We hope that those participating in Abu Dhabi as well as those joining remotely will leave the conference feeling newly inspired by the program and newly connected to our ever-growing community. Following EMNLP 2021 and major NLP conferences since, EMNLP 2022 is “hybrid,” serving both virtual and in-person participants.

Our key innovations for EMNLP 2022 include:

* EMNLP 2022 is “hybrid” in a second sense, as well: we allowed both direct and rolling review paper submissions, building on the pilot experiment of EMNLP 2021, which considered a small number of ARR submissions. 
* Familiar from NAACL but new to EMNLP, we’ve added an industry track.
* During the conference, “portals” will link virtual poster sessions to in-person conference participants during poster sessions each day.
* The first ACL-family conference in the United Arab Emirates.

 *Message from Noah A. Smith, University of Washington and Allen Institute for AI, Seattle, Washington, USA* 
***EMNLP 2022 General Chair***
 
[![](https://assets.underline.io/uploads/markdown_image/1/image/9eec7d4a287ee18c278b08229290aa83.png)](https://drive.google.com/file/d/1OlPv6QBeo62VVTughj2jkiLeyHd1WnUt/view)
 
[![](https://assets.underline.io/uploads/markdown_image/1/image/a3db7a768409f05192210d98601edb25.png)](https://emnlp2022.rocket.chat/)

To access this site you need to register. Please register [here](https://2022.emnlp.org/registration/).

Register here

Welcome!
EMNLP 2022 will take place in Abu Dhabi from December 7th to December 11th, 2022. And it will be held in hybrid mode, both online and offline.

Word associations are widely used in psychology to provide insights on how humans perceive and understand concepts. 
Comparing word associations in language models (LMs) to those generated by human subjects can serve as a proxy to uncover embedded lexical and commonsense knowledge in language models. 
While much helpful work has been done applying direct metrics, such as cosine similarity, to help understand latent spaces, these metrics are symmetric, while human word associativity is asymmetric. 
We propose WordTies, an algorithm based on constrained sampling from LMs, which allows an asymmetric measurement of associated words, given a cue word as the input. 
Comparing to existing methods, word associations found by this method share more overlap with associations provided by humans, and observe the asymmetric property of human associations. 
To examine possible reasons behind associations, we analyze the knowledge and reasoning behind the word pairings as they are linked to lexical and commonsense knowledge graphs.
When the knowledge about the nature of the word pairings is combined with a probability that the LM has learned that information, we have a new way to examine what information is captured in LMs.

WordTies: Measuring Word Associations in Language Models via Constrained Sampling

We conduct a large empirical evaluation to investigate the landscape of distributional robustness in question answering. Our investigation spans over 350 models and 16 question answering datasets, including a diverse set of architectures, model sizes, and adaptation methods (e.g., fine-tuning, adapter tuning, in-context learning, etc.). We find that, in many cases, model variations do not affect robustness and in-distribution performance alone determines out-of-distribution performance.
Moreover, our findings indicate that
i) zero-shot and in-context learning methods are more robust to distribution shifts than fully fine-tuned models;
ii) few-shot prompt fine-tuned models exhibit better robustness than few-shot fine-tuned span prediction models;
iii) parameter-efficient and robustness enhancing training methods provide no significant robustness improvements.
In addition, we publicly release all evaluations to encourage researchers to further analyze robustness trends for question answering models.

Exploring The Landscape of Distributional Robustness for Question Answering Models

Training state-of-the-art ASR systems such as RNN-T often has a high associated financial and environmental cost. Training with a subset of training data could mitigate this problem if the subset selected could achieve on-par performance with training with the entire dataset. Although there are many data subset selection(DSS) algorithms, direct application to the RNN-T is difficult, especially the DSS algorithms that are adaptive and use learning dynamics such as gradients, as RNN-T tend to have gradients with a significantly larger memory footprint. In this paper, we propose Partitioned Gradient Matching (PGM) a novel distributable DSS algorithm, suitable for massive datasets like those used to train RNN-T. Through extensive experiments on Librispeech 100H and Librispeech 960H, we show that PGM achieves between 3x to 6x speedup with only a very small accuracy degradation (under 1% absolute WER difference). In addition, we demonstrate similar results for PGM even in settings where the training data is corrupted with noise.

Partitioned Gradient Matching-based Data Subset Selection for Compute-Efficient Robust ASR Training

We present a robust methodology for evaluating biases in natural language generation(NLG) systems. Previous works use fixed hand-crafted prefix templates with mentions of various demographic groups to prompt models to generate continuations for bias analysis. These fixed prefix templates could themselves be specific in terms of styles or linguistic structures, which may lead to unreliable fairness conclusions that are not representative of the general trends from tone varying prompts. To study this problem, we paraphrase the prompts with different syntactic structures and use these to evaluate demographic bias in NLG systems. Our results suggest similar overall bias trends but some syntactic structures lead to contradictory conclusions compared to past works. We show that our methodology is more robust and that some syntactic structures prompt more toxic content while others could prompt less biased generation. This suggests the importance of not relying on a fixed syntactic structure and using tone-invariant prompts. Introducing syntactically-diverse prompts can achieve more robust NLG (bias) evaluation.

Towards Robust NLG Bias Evaluation with Syntactically-diverse Prompts

This work introduces a natural language inference (NLI) dataset that focuses on the validity of statements in legal wills. This dataset is unique because: (a) each entailment decision requires three inputs: the statement from the will, the law, and the conditions that hold at the time of the testator's death; and (b) the included texts are longer than the ones in current NLI datasets. We trained eight neural NLI models in this dataset. All the models achieve more than 80% macro F1 and accuracy, which indicates that neural approaches can handle this task reasonably well. However, group accuracy, a stricter evaluation measure that is calculated with a group of positive and negative examples generated from the same statement as a unit, is in mid 80s at best, which suggests that the models' understanding of the task remains superficial. Further ablative analyses and explanation experiments indicate that all three text segments are used for prediction, but some decisions rely on semantically irrelevant tokens. This indicates that overfitting on these longer texts likely happens, and that additional research is required for this task to be solved.

Validity Assessment of Legal Will Statements as Natural Language Inference

Currently, researchers focus on generating codes from the requirement documents. However, current approaches still perform poorly on some requirements needing complex problem-solving skills. In reality, to tackle such complex requirements, instead of directly translating requirement documents into codes, software engineers write codes via unified modeling language diagrams, such as flowcharts, an intermediate tool to analyze and visualize the system. Therefore, we propose a new source code generation task, that is, to generate source code from flowcharts with texts. We manually construct a benchmark dataset containing 320 flowcharts with their corresponding source codes. Obviously, it is not straightforward to employ the current approaches for the new source code generation task since (1) the flowchart is a graph that contains various structures, including loop, selection, and others which is different from texts; (2) the connections between nodes in the flowchart are abundant and diverse which need to be carefully handled. To solve the above problems, we propose a two-stage code generation model. In the first stage, a structure recognition algorithm is employed to transform the flowchart into pseudo-code containing the structural conventions of a typical programming language such as while, if. In the second stage, a code generation model is employed to convert the pseudo-code into code. Experimental results show that the proposed approach can achieve some improvement over the baselines.

Code Generation From Flowcharts with Texts: A Benchmark Dataset and An Approach

Syntactic trees have been widely applied in relation extraction (RE). However, since parsing qualities are not stable on different text domains and a pre-defined grammar may not well fit the target relation schema, the introduction of syntactic structures sometimes fails to improve RE performances consistently. In this work, we study RE models with various unsupervised structures mined from pre-trained language models (e.g., BERT). We show that, similar to syntactic trees, unsupervised structures are quite informative for RE task: they are able to obtain competitive (even the best) performance scores on benchmark RE datasets (ACE05, WebNLG, SciERC). We also conduct detailed analyses on their abilities of adapting new RE domains and influence of noise links in those structures. The results suggest that unsupervised structures are reasonable alternatives of commonly used syntactic structures in relation extraction models.

Explore Unsupervised Structures in Pretrained Models for Relation Extraction

Adversarial examples in NLP are receiving increasing research attention. One line of investigation is the generation of word-level adversarial examples against fine-tuned Transformer models that preserve naturalness and grammaticality. Previous work found that human- and machine-generated adversarial examples are comparable in their naturalness and grammatical correctness. Most notably, humans were able to generate adversarial examples much more effortlessly than automated attacks. In this paper, we provide a detailed analysis of exactly how humans create these adversarial examples. By exploring the behavioural patterns of human workers during the generation process, we identify statistically significant tendencies based on which words humans prefer to select for adversarial replacement (e.g., word frequencies, word saliencies, sentiment) as well as where and when words are replaced in an input sequence. With our findings, we seek to inspire efforts that harness human strategies for more robust NLP models.

Identifying Human Strategies for Generating Word-Level Adversarial Examples

Zero-Shot Learning (ZSL) has shown great promise at the intersection of vision and language, and generative methods for ZSL are predominant owing to their efficiency. Moreover, textual description or attribute plays a critical role in transferring knowledge from the seen to unseen classes in ZSL. Such generative approaches for ZSL are very costly to train and require the class description of the unseen classes during training. In this work, we propose a non-generative gating-based attribute refinement network for ZSL, which achieves similar accuracies to generative methods of ZSL, at a much lower computational cost. The refined attributes are mapped into the visual domain through an attribute embedder, and the whole network is guided by the circle loss and the well-known softmax cross-entropy loss to obtain a robust class embedding. We refer to our approach as Circle loss guided gating-based Attribute-Refinement Network (CARNet). We perform extensive experiments on the five benchmark datasets over the various challenging scenarios viz., Generalized ZSL (GZSL), Continual GZSL (CGZSL), and conventional ZSL. We observe that the CARNet significantly outperforms recent non-generative ZSL methods and most generative ZSL methods in all three settings by a significant margin. Our extensive ablation study disentangles the performance of various components and justifies their importance. The source code is available at https://github.com/Sethup123/CARNet.

Refinement Matters: Textual Description Needs to be Refined for Zero-shot Learning

Discovering new slots is critical to the success of dialogue systems. Most existing methods rely on automatic slot induction in unsupervised fashion or perform domain adaptation across zero or few-shot scenarios. They have difficulties in providing high-quality supervised signals to learn clustering-friendly features, and are limited in effectively transferring the prior knowledge from known slots to new slots. In this work, we propose a Semi-supervised Incremental Clustering method (SIC), to discover new slots with the aid of existing linguistic annotation models and limited known slot data. Specifically, we harvest slot value candidates with NLP model cues and innovatively formulate the slot discovery task under an incremental clustering framework. The model gradually calibrate slot representations under the supervision of generated pseudo-labels, and automatically learns to terminate when no more salient slot remains. Our thorough evaluation on five public datasets demonstrates that it significantly outperforms state-of-the-art models.

Downloads

Next from EMNLP 2022

WordTies: Measuring Word Associations in Language Models via Constrained Sampling

Similar lecture

InterFair: Debiasing with Natural Language Feedback for Fair Interpretable Predictions

.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 2022

WordTies: Measuring Word Associations in Language Models via Constrained Sampling

Similar lecture

InterFair: Debiasing with Natural Language Feedback for Fair Interpretable Predictions

Downloads