2012
DOI: 10.1007/978-3-642-35292-8_16
|View full text |Cite
|
Sign up to set email alerts
|

Mutual Information and Perplexity Based Clustering of Dialogue Information for Dynamic Adaptation of Language Models

Abstract: Abstract. We present two approaches to cluster dialogue-based information obtained by the speech understanding module and the dialogue manager of a spoken dialogue system. The purpose is to estimate a language model related to each cluster, and use them to dynamically modify the model of the speech recognizer at each dialogue turn. In the first approach we build the cluster tree using local decisions based on a Maximum Normalized Mutual Information criterion. In the second one we take global decisions, based o… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1

Citation Types

0
2
0

Year Published

2020
2020
2020
2020

Publication Types

Select...
2

Relationship

0
2

Authors

Journals

citations
Cited by 2 publications
(2 citation statements)
references
References 8 publications
0
2
0
Order By: Relevance
“…Different metrics such as potential (sum of squared distances of samples to their closest cluster center) [12], log-likelihood score [12], perplexity score (information measure of generative probabilistic models) [18], AIC [12], and BIC [12], are used to assess clustering quality based on model assumption or information theory. However, they are not able to assess the stability of a clustering.…”
Section: A Lemma For Swapping Lines For Nmfmentioning
confidence: 99%
“…Different metrics such as potential (sum of squared distances of samples to their closest cluster center) [12], log-likelihood score [12], perplexity score (information measure of generative probabilistic models) [18], AIC [12], and BIC [12], are used to assess clustering quality based on model assumption or information theory. However, they are not able to assess the stability of a clustering.…”
Section: A Lemma For Swapping Lines For Nmfmentioning
confidence: 99%
“…As regards the performance of the system, Table 8.16 shows the improvements on the recognition, understanding and dialogue management metrics for the NMI-based and the minimum global perplexity based clustering strategies (Lucas-Cuesta et al, 2012).…”
Section: Experimental Frameworkmentioning
confidence: 99%