Defending Against Neural Fake News

NeurIPS 2019 Rowan ZellersAri HoltzmanHannah RashkinYonatan BiskAli FarhadiFranziska RoesnerYejin Choi

Recent progress in natural language generation has raised dual-use concerns. While applications like summarization and translation are positive, the underlying technology also might enable adversaries to generate neural fake news: targeted propaganda that closely mimics the style of real news... (read more)

PDF Abstract

Evaluation Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK COMPARE
Fake News Detection Grover-Mega GPT2 (355M) Unpaired Accuracy 70.1% # 4
Fake News Detection Grover-Mega Grover-Mega Unpaired Accuracy 92.0% # 1
Fake News Detection Grover-Mega Grover-Large Unpaired Accuracy 80.8% # 2
Fake News Detection Grover-Mega BERT-Large Unpaired Accuracy 73.1% # 3