Uncertainty Estimation in Autoregressive Structured Prediction
This addresses the problem of safety and robustness in AI systems for researchers and practitioners in NLP and speech processing, but it is incremental as it extends existing uncertainty estimation methods to structured prediction.
The paper tackles uncertainty estimation for autoregressive structured prediction tasks, proposing a unified probabilistic ensemble-based framework and providing baselines for error and out-of-domain detection on translation and speech recognition datasets.
Uncertainty estimation is important for ensuring safety and robustness of AI systems. While most research in the area has focused on un-structured prediction tasks, limited work has investigated general uncertainty estimation approaches for structured prediction. Thus, this work aims to investigate uncertainty estimation for autoregressive structured prediction tasks within a single unified and interpretable probabilistic ensemble-based framework. We consider: uncertainty estimation for sequence data at the token-level and complete sequence-level; interpretations for, and applications of, various measures of uncertainty; and discuss both the theoretical and practical challenges associated with obtaining them. This work also provides baselines for token-level and sequence-level error detection, and sequence-level out-of-domain input detection on the WMT'14 English-French and WMT'17 English-German translation and LibriSpeech speech recognition datasets.