Weakly Supervised Natural Language Learning Without Redundant Views

We investigate single-view algorithms as an alternative to multi-view algorithms for weakly supervised learning for natural language processing tasks without a natural feature split. In particular, we apply co-training, self-training, and EM to one such task and find that both selftraining and FS-EM, a new variation of EM that incorporates feature selection, outperform cotraining and are comparatively less sensitive to parameter changes.

• Type: Conference
• Year: 2003
• Where: NAACL
• Authors: Vincent Ng, Claire Cardie
• Keywords: Multi-view Algorithms, NAACL 2003, NAACL 2007, Natural Feature Split, Natural Language Processing