Share on Facebook Tweet on Twitter Share on LinkedIn Share by email
Word Confidence Calibration Using a Maximum Entropy Model with Constraints on Confidence and Word Distributions

Dong Yu, Shizhen Wang, Jinyu Li, and Li Deng

Abstract

It is widely known that the quality of confidence measure is critical for speech applications. In this paper, we present our recent work on improving word confidence scores by calibrating them using a small set of calibration data when only the recognized word sequence and associated raw confidence scores are made available. The core of our technique is the maximum entropy model with distribution constraints which naturally and effectively make use of the word distribution, the raw confidence-score distribution, and the context information. We demonstrate the effectiveness of our approach by showing that it can achieve relative 38% mean square error (MSE), 39% negative normalized likelihood (NNLL), and 23% equal error rate (EER) reduction on a voice mail transcription data set and relative 35% MSE, 45% NNLL, and 35% EER reduction on a command and control data set.

Details

Publication typeInproceedings
PublisherIEEE
> Publications > Word Confidence Calibration Using a Maximum Entropy Model with Constraints on Confidence and Word Distributions