Canada

Recent task-oriented dialog systems have had great success in building English-based personal assistants, but extending these systems to a global audience is challenging due to the need for annotated data in the target language. An alternative approach is to leverage existing data in a high-resource language to enable cross-lingual transfer in low-resource language models. However, this type of transfer has not been widely explored in natural language response generation. In this research, we investigate the use of state-of-the-art multilingual models such as mBART and T5 to facilitate zero-shot and few-shot transfer of code-switched responses. We propose a new adapter-based framework that allows for efficient transfer by learning task-specific representations and encapsulating source and target language representations. Our framework is able to successfully transfer language knowledge even when the target language corpus is limited. We present both quantitative and qualitative analyses to evaluate the effectiveness of our approach.

&lt;iframe src=&quot;https://app.sli.do/event/or9aypsi961h595NQtD7kk/embed/polls/d244e08b-75dd-4f38-8d94-d37eaa016754&quot; width=&quot;300&quot; height=&quot;400&quot;&gt;&lt;/iframe&gt;

ACL 2023

Towards Zero-Shot Multilingual Transfer for Code-Switched Responses

multilingual dialog

code-switched

response generation

few-shot

language models

Recent task-oriented dialog systems have had great success in building English-based personal assistants, but extending these systems to a global audience is challenging due to the need for annotated data in the target language. An alternative approach is to leverage existing data in a high-resource language to enable cross-lingual transfer in low-resource language models. However, this type of transfer has not been widely explored in natural language response generation. In this research, we investigate the use of state-of-the-art multilingual models such as mBART and T5 to facilitate zero-shot and few-shot transfer of code-switched responses. We propose a new adapter-based framework that allows for efficient transfer by learning task-specific representations and encapsulating source and target language representations. Our framework is able to successfully transfer language knowledge even when the target language corpus is limited. We present both quantitative and qualitative analyses to evaluate the effectiveness of our approach.

<iframe src="https://app.sli.do/event/or9aypsi961h595NQtD7kk/embed/polls/d244e08b-75dd-4f38-8d94-d37eaa016754" width="300" height="400"></iframe>

technical paper

### Welcome to ACL 2023, the 61st Annual Meeting of the Association for Computational Linguistics! 
<br>The conference will be held in Toronto, Canada, July 9-14, 2023.<br><br>
Following the succession of the recent conferences in our field, ACL 2023 will adopt a hybrid format.
While the impact of Covid has considerably diminished in terms of traveling, obtaining visas to Canada
entails a very long process. Moreover, the global economic conditions pose challenges for many individuals to travel to conferences. Recognizing these circumstances, we know many participants may not be
able to attend the conference in person. Therefore, we are committed to providing a great virtual platform
so everyone has the opportunity to interact with other participants and enjoy the conference. Based on the
current registered participants, approxiately 30% have chosen to attend the conference virtually. Whether
you join us in person or virtually, we sincerely hope everyone has a remarkable conference experience.<br>
This General Chair’s message is where I express my gratitude to the many individuals who have made
enormous contributions to the conference over the past year.

Read [**ACL 2023 General Chair's message**](https://docs.google.com/document/d/1WobYM7norbG4dI48s75HfJoD89qgX5a_F-6U8AteLSA/edit?usp=sharing/) in full.

##### **[Conference Handbook](https://2023.aclweb.org/downloads/acl2023-handbook.pdf)**
<br><br><br><br>

The Association for Computational Linguistics (ACL) is the premier international scientific and professional society for people working on computational problems involving human language, a field often referred to as either computational linguistics or natural language processing.

Despite remarkable progress made in natural language processing, even the state-of-the-art models often make incorrect predictions. Such predictions hamper the reliability of systems and limit their widespread adoption in real-world applications. 'Selective prediction' partly addresses the above concern by enabling models to abstain from answering when their predictions are likely to be incorrect. While selective prediction is advantageous, it leaves us with a pertinent question 'what to do after abstention'. To this end, we present an explorative study on 'Post-Abstention', a task that allows re-attempting the abstained instances with the aim of increasing **coverage** of the system without significantly sacrificing its **accuracy**. We first provide mathematical formulation of this task and then explore several methods to solve it. Comprehensive experiments on 11 QA datasets show that these methods lead to considerable risk improvements --performance metric of the Post-Abstention task-- both in the in-domain and the out-of-domain settings. We also conduct a thorough analysis of these results which further leads to several interesting findings. Finally, we believe that our work will encourage and facilitate further research in this important area of addressing the reliability of NLP systems.

<iframe src="https://app.sli.do/event/8PaBoicsuHK7beXFunvrtr/embed/polls/3a865239-1045-45dd-937e-d684372f41cd" width="300" height="400"></iframe>

Post-Abstention: Towards Reliably Re-Attempting the Abstained Instances in QA

This evidence-based position paper critiques current research practices within the language model pre-training literature. Despite rapid recent progress afforded by increasingly better pre-trained language models (PLMs), current PLM research practices often conflate different possible sources of model improvement, without conducting proper ablation studies and principled comparisons between different models under comparable conditions. These practices (i) leave us ill-equipped to understand which pre-training approaches should be used under what circumstances; (ii) impede reproducibility and credit assignment; and (iii) render it difficult to understand: "How exactly does each factor contribute to the progress that we have today?'' We provide a case in point by revisiting the success of BERT over its baselines, ELMo and GPT-1, and demonstrate how --- under comparable conditions where the baselines are tuned to a similar extent --- these baselines (and even-simpler variants thereof) can, in fact, achieve competitive or better performance than BERT. These findings demonstrate how disentangling different factors of model improvements can lead to valuable new insights. We conclude with recommendations for how to encourage and incentivize this line of work, and accelerate progress towards a better and more systematic understanding of what factors drive the progress of our foundation models today.

<iframe src="https://app.sli.do/event/or9aypsi961h595NQtD7kk/embed/polls/628808b4-2c36-4b77-b3a0-497f59b2749b" width="300" height="400"></iframe>

On "Scientific Debt" in NLP: A Case for More Rigour in Language Model Pre-Training Research

Recent advances in open-domain question answering (ODQA) have demonstrated impressive accuracy on general-purpose domains like Wikipedia. While some work has been investigating how well ODQA models perform when tested for out-of-domain (OOD) generalization, these studies have been conducted only under conservative shifts in data distribution and typically focus on a single component (i.e., retriever or reader) rather than an end-to-end system. This work proposes a more realistic end-to-end domain shift evaluation setting covering five diverse domains. We not only find that end-to-end models fail to generalize but that high retrieval scores often still yield poor answer prediction accuracy. To address these failures, we investigate several interventions, in the form of data augmentations, for improving model adaption and use our evaluation set to elucidate the relationship between the efficacy of an intervention scheme and the particular type of dataset shifts we consider. We propose a generalizability test that estimates the type of shift in a target dataset without training a model in the target domain and that the type of shift is predictive of which data augmentation schemes will be effective for domain adaption. Overall, we find that these interventions increase end-to-end performance by up to ~24 points.

<iframe src="https://app.sli.do/event/oApzFnCrRo86e5qieE2QHQ/embed/polls/5f2202c0-576f-40e1-b5b1-d5dc54420c5a" width="300" height="400"></iframe>

To Adapt or to Annotate: Challenges and Interventions for Domain Adaptation in Open-Domain Question Answering

Compositional generalization allows efficient learning and human-like inductive biases. Since most research investigating compositional generalization in NLP is done on English, important questions remain underexplored. Do the necessary compositional generalization abilities differ across languages? Can models compositionally generalize cross-lingually? As a first step to answering these questions, recent work used neural machine translation to translate datasets for evaluating compositional generalization in semantic parsing. However, we show that this entails critical semantic distortion. To address this limitation, we craft a faithful rule-based translation of the MCWQ dataset from English to Chinese and Japanese. Even with the resulting robust benchmark, which we call MCWQ-R, we show that the distribution of compositions still suffers due to linguistic divergences, and that multilingual models still struggle with cross-lingual compositional generalization. Our dataset and methodology will serve as useful resources for the study of cross-lingual compositional generalization in other tasks.

<iframe src="https://app.sli.do/event/eX4PwSmcCc1EvCqjxcYdKB/embed/polls/91a9a5be-0aac-4185-8d5c-990bc29d6353" width="300" height="400"></iframe>

On Evaluating Multilingual Compositional Generalization with Translated Datasets

Large language models (LLMs) that have been trained on multilingual but not parallel text exhibit a remarkable ability to translate between languages. We probe this ability in an in-depth study of the pathways language model (PaLM), which has demonstrated the strongest machine translation (MT) performance among similarly-trained LLMs to date. We investigate various strategies for choosing translation examples for few-shot prompting, concluding that example quality is the most important factor. Using optimized prompts, we revisit previous assessments of PaLM's MT capabilities with more recent test sets, modern MT metrics, and human evaluation, and find that its performance, while impressive, still lags that of state-of-the-art supervised systems. We conclude by providing an analysis of PaLM's MT output which reveals some interesting properties and prospects for future work.

<iframe src="https://app.sli.do/event/t38CPUax2NoHBQcXN6okdA/embed/polls/838c799e-2935-4d74-8dd7-0e12d01f6e49" width="300" height="400"></iframe>

Prompting PaLM for Translation: Assessing Strategies and Performance

Health-related speech datasets are often small and varied in focus. This makes it difficult to leverage them to effectively support healthcare goals. Robust transfer of linguistic features across different datasets orbiting the same goal carries potential to address this concern. To test this hypothesis, we experiment with domain adaptation (DA) techniques on heterogeneous spoken language data to evaluate generalizability across diverse datasets for a common task: dementia detection. We find that adapted models exhibit better performance across conversational and task-oriented datasets. The feature-augmented DA method achieves a 22% increase in accuracy adapting from a conversational to task-specific dataset compared to a jointly trained baseline. This suggests promising capacity of these techniques to allow for productive use of disparate data for a complex spoken language healthcare task.

<iframe src="https://app.sli.do/event/iHwnn5G9xYwcAmSbeKA4Au/embed/polls/dff53abe-71f6-4fd0-8418-bfc9577037db" width="300" height="400"></iframe>

Towards Domain-Agnostic and Domain-Adaptive Dementia Detection from Spoken Language

Weakly supervised learning is a popular approach for training machine learning models in low-resource settings. Instead of requesting high-quality yet costly human annotations, it allows training models with noisy annotations obtained from various weak sources. Recently, many sophisticated approaches have been proposed for robust training under label noise, reporting impressive results. In this paper, we revisit the setup of these approaches and find that the benefits brought by these approaches are significantly overestimated. Specifically, we find that the success of existing weakly supervised learning approaches heavily relies on the availability of clean validation samples which, as we show, can be leveraged much more efficiently by simply training on them. After using these clean labels in training, the advantages of using these sophisticated approaches are mostly wiped out. This remains true even when reducing the size of the available clean data to just five samples per class, making these approaches impractical. To understand the true value of weakly supervised learning, we thoroughly analyse diverse NLP datasets and tasks to ascertain when and why weakly supervised approaches work, and provide recommendations for future research.

<iframe src="https://app.sli.do/event/jQes3hFZo3TYjGXEHm6w95/embed/polls/80ca54dd-ad80-4a55-bd29-faa8c15a9307" width="300" height="400"></iframe>

Weaker Than You Think: A Critical Look at Weakly Supervised Learning

In-context learning (ICL) is an important paradigm for adapting large language models (LLMs) to new tasks, but the generalization behavior of ICL remains poorly understood. We investigate the inductive biases of ICL from the perspective of feature bias: which feature ICL is more likely to use given a set of underspecified demonstrations in which two features are equally predictive of the labels. First, we characterize the feature biases of GPT-3 models by constructing underspecified demonstrations from a range of NLP datasets and feature combinations. We find that LLMs exhibit clear feature biasesfor example, demonstrating a strong bias to predict labels according to sentiment rather than shallow lexical features, like punctuation. Second, we evaluate the effect of different interventions that are designed to impose an inductive bias in favor of a particular feature, such as adding a natural language instruction or using semantically relevant label words. We find that, while many interventions can influence the learner to prefer a particular feature, it can be difficult to overcome strong prior biases. Overall, our results provide a broader picture of the types of features that ICL may be more likely to exploit and how to impose inductive biases that are better aligned with the intended task.

<iframe src="https://app.sli.do/event/pk54NpNnxuhWHUa3vSMXH2/embed/polls/53d3406d-b457-40f2-a179-a4c85f9aeb6b" width="300" height="400"></iframe>

Measuring Inductive Biases of In-Context Learning with Underspecified Demonstrations

Large pre-trained language models achieve impressive results across many tasks. However, recent works point out that pre-trained language models may memorize a considerable fraction of their training data, leading to the privacy risk of information leakage. In this paper, we propose a method named Ethicist for targeted training data extraction through loss smoothed soft prompting and calibrated confidence estimation, investigating how to recover the suffix in the training data when given a prefix. To elicit memorization in the attacked model, we tune soft prompt embeddings while keeping the model fixed. We further propose a smoothing loss that smooths the loss distribution of the suffix tokens to make it easier to sample the correct suffix. In order to select the most probable suffix from a collection of sampled suffixes and estimate the prediction confidence, we propose a calibrated confidence estimation method, which normalizes the confidence of the generated suffixes with a local estimation. We show that Ethicist significantly improves the extraction performance on a recently proposed public benchmark. We also investigate several factors influencing the data extraction performance, including decoding strategy, model scale, prefix length, and suffix length. Our code is availabel at https://github.com/thu-coai/Targeted-Data-Extraction.

<iframe src="https://app.sli.do/event/1kgCXjyZMgdV89PsksaJsj/embed/polls/10813b4b-fed2-4c7e-8e27-d454ac5d465b" width="300" height="400"></iframe>

ETHICIST: Targeted Training Data Extraction Through Loss Smoothed Soft Prompting and Calibrated Confidence Estimation

We propose a novel approach to automatically generate distractors for cloze exercises for English language learners, using round-trip neural machine translation. A carrier sentence is translated from English into another (pivot) language and back, and distractors are produced by aligning the original sentence with its round-trip translation. We make use of 16 linguistically-diverse pivots and generate hundreds of translation hypotheses in each direction. We show that using hundreds of translations allows us to generate a rich set of challenging distractors. Moreover, we find that typologically unrelated language pivots contribute more diverse candidate distractors, compared to language pivots that are closely related. We further evaluate the use of machine translation systems of varying quality and find that better quality MT systems produce more challenging distractors. Finally, we conduct a study with language learners, demonstrating that the automatically generated distractors are of the same difficulty as the gold distractors produced by human experts.

<iframe src="https://app.sli.do/event/qQ87ZDnDYN6cfmCQWHFaok/embed/polls/3a2adf93-0bdf-4105-8059-19ad796d136e" width="300" height="400"></iframe>

Premium content

Towards Zero-Shot Multilingual Transfer for Code-Switched Responses

Downloads

Next from ACL 2023

Post-Abstention: Towards Reliably Re-Attempting the Abstained Instances in QA

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES