<?xml version="1.0" standalone="yes"?>
<Paper uid="H94-1013">
  <Title>Weide, R., Huang, X., and Alleva, F., &amp;quot;Improving Speech- Recognition Performance Via Phone-Dependent VQ Codebooks, Multiple Speaker Clusters And Adaptive Language Models&amp;quot;, ARPA Spoken Language Systems Workshop, March</Title>
  <Section position="10" start_page="79" end_page="80" type="concl">
    <SectionTitle>
8. SUMMARY
</SectionTitle>
    <Paragraph position="0"> We described our latest attempt at adaptive language modeling. At the heart of our approach is a Maximum Entropy (ME) model, which incorporates many knowledge sources in a consistent manner. We have demonstrated that the ME model significantly improves on the conventional static trigram, a challenge which has evaded many past attempts(\[17, 18\]).</Paragraph>
    <Paragraph position="1"> The approach is particularly applicable in domains with a modest amount of LM training data.</Paragraph>
    <Paragraph position="2">  interpolated ad_~ptive models, for both eross-domain and fimited-data adaptation, testing on 420KW of unseen AP wire</Paragraph>
  </Section>
class="xml-element"></Paper>