Unsupervised Selective Rationalization with Noise Injection

Citations Over TimeTop 24% of 2023 papers

Abstract

A major issue with using deep learning models in sensitive applications is that they provide no explanation for their output. To address this problem, unsupervised selective rationalization produces rationales alongside predictions by chaining two jointly-trained components, a rationale generator and a predictor. Although this architecture guarantees that the prediction relies solely on the rationale, it does not ensure that the rationale contains a plausible explanation for the prediction. We introduce a novel training technique that effectively limits generation of implausible rationales by injecting noise between the generator and the predictor. Furthermore, we propose a new benchmark for evaluating unsupervised selective rationalization models using movie reviews from existing datasets. We achieve sizeable improvements in rationale plausibility and task accuracy over the state-of-the-art across a variety of tasks, including our new benchmark, while maintaining or improving model faithfulness.

Related Papers

• → A Tour of Unsupervised Deep Learning for Medical Image Analysis(2021)94 cited
• → Introduction to supervised machine learning in clinical epidemiology(2022)26 cited
• → Application of Machine Learning in Animal Disease Analysis and Prediction(2020)26 cited
• → Advances in Machine Learning and Deep Learning Algorithms and Their Applications(2021)1 cited
• → Survey of Machine Learning Algorithms & its Applications(2021)1 cited