Learning The Difference That Makes A Difference With Counterfactually-Augmented Data

Divyansh Kaushik
Divyansh Kaushik

ICLR, 2020.

Cited by: 93|Views488
EI
Weibo:
Our methods appear useful on both sentiment analysis and Natural Language Inference, two contrasting tasks

Abstract:

Despite alarm over the reliance of machine learning systems on so-called spurious patterns, the term lacks coherent meaning in standard statistical frameworks. However, the language of causality offers clarity: spurious associations are due to confounding (e.g., a common cause), but not direct or indirect causal effects. In this paper, we...More
0
Your rating :
0

 

Tags
Comments