Paper: Information-theoretic Multi-view Domain Adaptation

ACL ID P12-2053
Title Information-theoretic Multi-view Domain Adaptation
Venue Annual Meeting of the Association of Computational Linguistics
Session Short Paper
Year 2012
Authors

We use multiple views for cross-domain doc- ument classification. The main idea is to strengthen the views? consistency for target data with source training data by identify- ing the correlations of domain-specific fea- tures from different domains. We present an Information-theoretic Multi-view Adapta- tion Model (IMAM) based on a multi-way clustering scheme, where word and link clus- ters can draw together seemingly unrelated domain-specific features from both sides and iteratively boost the consistency between doc- ument clusterings based on word and link views. Experiments show that IMAM signifi- cantly outperforms state-of-the-art baselines.