Explain and Predict, and then Predict Again
This work is significant for researchers and practitioners in natural language processing who aim to build more transparent and effective AI systems, offering an incremental improvement in integrating rationales data for better task performance.
This paper addresses the challenge of creating learning systems that are both effective and interpretable. The authors propose ExPred, a novel multi-task learning approach that first generates an extractive explanation from text, then uses a prediction network on these explanations to optimize task performance. ExPred substantially outperforms existing approaches on fact verification, sentiment classification, and QA datasets.
A desirable property of learning systems is to be both effective and interpretable. Towards this goal, recent models have been proposed that first generate an extractive explanation from the input text and then generate a prediction on just the explanation called explain-then-predict models. These models primarily consider the task input as a supervision signal in learning an extractive explanation and do not effectively integrate rationales data as an additional inductive bias to improve task performance. We propose a novel yet simple approach ExPred, that uses multi-task learning in the explanation generation phase effectively trading-off explanation and prediction losses. And then we use another prediction network on just the extracted explanations for optimizing the task performance. We conduct an extensive evaluation of our approach on three diverse language datasets -- fact verification, sentiment classification, and QA -- and find that we substantially outperform existing approaches.