Paper-Conference

GL-CLeF: A Global--Local Contrastive Learning Framework for Cross-lingual Spoken Language Understanding
A contrastive cross-lingual SLU framework with global and local alignment components.
Cite DOI URL
GL-CLeF: A Global--Local Contrastive Learning Framework for Cross-lingual Spoken Language Understanding
Interpreting the Robustness of Neural NLP Models to Textual Perturbations
A learnability-based explanation of why NLP models are brittle to textual perturbations.
Cite DOI URL
Interpreting the Robustness of Neural NLP Models to Textual Perturbations
N-Shot Learning for Augmenting Task-Oriented Dialogue State Tracking
A belief-state-driven dialogue augmentation method that improves n-shot task-oriented DST.
Cite DOI URL
N-Shot Learning for Augmenting Task-Oriented Dialogue State Tracking
So Different Yet So Alike! Constrained Unsupervised Text Style Transfer
A constrained unsupervised text style transfer method with cooperative losses for preserving content and attributes.
Cite DOI URL
So Different Yet So Alike! Constrained Unsupervised Text Style Transfer
Reliability Testing for Natural Language Processing Systems
A framework for reliability testing of NLP systems in fairness, robustness, and transparency settings.
Cite DOI URL
Reliability Testing for Natural Language Processing Systems
Zero-shot Fact Verification by Claim Generation
A claim-generation framework that improves zero-shot fact verification with synthetic training data.
Cite DOI URL
Zero-shot Fact Verification by Claim Generation
Velocidapter: Task-oriented Dialogue Comprehension Modeling Pairing Synthetic Text Generation with Domain Adaptation
A synthetic dialogue generation framework that improves low-resource task-oriented dialogue comprehension.
Cite DOI URL
Velocidapter: Task-oriented Dialogue Comprehension Modeling Pairing Synthetic Text Generation with Domain Adaptation
Unsupervised Multi-hop Question Answering by Question Generation
An unsupervised multi-hop QA framework that generates synthetic training data from heterogeneous sources.
Cite DOI URL
Unsupervised Multi-hop Question Answering by Question Generation
Analyzing the Domain Robustness of Pretrained Language Models, Layer by Layer
A layer-by-layer study of how pretrained language model representations vary across domains.
Cite URL
Analyzing the Domain Robustness of Pretrained Language Models, Layer by Layer
Exploring Question-Specific Rewards for Generating Deep Questions
Recent question generation (QG) approaches often utilize the sequence-to-sequence framework (Seq2Seq) to optimize the log likelihood of …
Cite DOI URL