Modeling Confidence in Sequence-to-Sequence Models

TL;DR

This paper introduces methods to assess the confidence of sequence-to-sequence models by measuring similarity between training and test conditions, improving confidence estimation for tasks like machine translation and speech recognition.

Contribution

It proposes novel confidence estimation techniques based on similarity measures, including internal alignment models, applicable across multiple sequence-to-sequence tasks.

Findings

Improved segment-level confidence estimation in machine translation.

Enhanced confidence annotation for source tokens.

Identified 60% of errors by examining 20% of speech recognition data.

Abstract

Recently, significant improvements have been achieved in various natural language processing tasks using neural sequence-to-sequence models. While aiming for the best generation quality is important, ultimately it is also necessary to develop models that can assess the quality of their output. In this work, we propose to use the similarity between training and test conditions as a measure for models' confidence. We investigate methods solely using the similarity as well as methods combining it with the posterior probability. While traditionally only target tokens are annotated with confidence measures, we also investigate methods to annotate source tokens with confidence. By learning an internal alignment model, we can significantly improve confidence projection over using state-of-the-art external alignment tools. We evaluate the proposed methods on downstream confidence estimation for machine translation (MT). We show improvements on segment-level confidence estimation as well as on confidence estimation for source tokens. In addition, we show that the same methods can also be applied to other tasks using sequence-to-sequence models. On the automatic speech recognition (ASR) task, we are able to find 60% of the errors by looking at 20% of the data.