conftrace_

2019 EMNLP EMNLP 2019

Domain-agnostic Question-Answering with Adversarial Training

Abstract

AbstractAdapting models to new domain without finetuning is a challenging problem in deep learning. In this paper, we utilize an adversarial training framework for domain generalization in Question Answering (QA) task. Our model consists of a conventional QA model and a discriminator. The training is performed in the adversarial manner, where the two models constantly compete, so that QA model can learn domain-invariant features. We apply this approach in MRQA Shared Task 2019 and show better performance compared to the baseline model.

🌱 Topic Pioneer - Domain Generalization

🌉 Interdisciplinary Bridge - Artificial Intelligence and Natural Language Processing

📈 Trend Setter - Domain Generalization

Authors

Seanie Lee, Donggyu Kim, Jangwon Park

Topics

Machine Learning > Learning Types > Adversarial Learning Machine Learning > Application Areas > Domain Generalization Natural Language Processing > Applications > Question Answering Artificial Intelligence > Core AI > Adversarial Learning Deep Learning > Learning Types > Adversarial Learning Artificial Intelligence > Learning Paradigms > Domain Generalization

Keywords

adversarial learning feature learning domain generalization domain adaptation feature extraction question answering adversarial training deep learning domain-invariant feature

More from EMNLP 2019

All EMNLP 2019 papers →

Extractive NarrativeQA with Heuristic Pre-Training

CLER: Cross-task Learning with Expert Representation to Generalize Reading and Understanding

Question Answering Using Hierarchical Attention on Top of BERT Features

Generalizing Question Answering System with Pre-trained Language Model Fine-tuning

D-NET: A Pre-Training and Fine-Tuning Framework for Improving the Generalization of Machine Reading Comprehension

An Exploration of Data Augmentation and Sampling Techniques for Domain-Agnostic Question Answering