Robust cross-domain disfluency detection with pattern match networks

Abstract

In this paper we introduce a novel pattern match neural network architecture that uses neighbor similarity scores as features, eliminating the need for feature engineering in a disfluency detection task. We evaluate the approach in disfluency detection for four different speech genres, showing that the approach is as effective as hand-engineered pattern match features when used on in-domain data and achieves superior performance in cross-domain scenarios.

This paper was submitted to EMNLP 2018 and was rejected. Our EMNLP submission is posted here to establish concurrency with "Disfluency Detection using Auto-Correlational Neural Networks" by P. Lou, P. Anderson, M. Johnson which was submitted to EMNLP at the same time

References in corpus (1)

• Parsing Speech: A Neural Approach to Integrating Lexical and Acoustic-Prosodic Information

Cited by in corpus (1)

• Combining Self-Training and Self-Supervised Learning for Unsupervised Disfluency Detection