Work

Trick Me If You Can: Human-in-the-Loop Generation of Adversarial Examples for Question Answering

Year: 2019

Type: article

Abstract: Adversarial evaluation stress-tests a model’s understanding of natural language. Because past approaches expose superficial patterns, the resulting adversarial examples are limited in complexity and diversity. We propose human- in-the-loop adversarial generation, where human authors are guided to break models. We aid the authors with interpretations of model predictions through an interactive user interface. We apply this generation framework to a question answering task called Quizbowl, where trivia enthusiasts craft adversarial questions. The resulting questions are validated via live human–computer matches: Although the questions appear ordinary to humans, they systematically stump neural and information retrieval models. The adversarial questions cover diverse phenomena from multi-hop reasoning to entity type distractors, exposing open challenges in robust question answering.

Source: Transactions of the Association for Computational Linguistics

DOI: 10.1162/tacl_a_00279

Authors: Eric Wallace, Pedro Rodriguez, Shi Feng, Ikuya Yamada, Jordan Boyd-Graber

Institution: University of Maryland, College Park

Language: English

Topic: Adversarial Robustness in Machine Learning

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: diamond

Funders: National Science Foundation, Raytheon Company, Defense Advanced Research Projects Agency

Awards: CAREER: Human-Computer Cooperation for Word-by-Word Question Answering, G211012689, G2515541506, III: Medium: Collaborative Research: Closing the User-Model Loop for Understanding Topics in Large Document Collections