A data-selection method that expands from confirmed assistant errors via k-nearest neighbors in BiLSTM embedding spaces outperforms random and entropy-based active learning for domain classification.
Synthesis Lectures on Artificial Intel- ligence and Machine Learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2019 1verdicts
UNVERDICTED 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
Active Learning for Domain Classification in a Commercial Spoken Personal Assistant
A data-selection method that expands from confirmed assistant errors via k-nearest neighbors in BiLSTM embedding spaces outperforms random and entropy-based active learning for domain classification.