Class Meeting for 10-707 10/5/2009

This is one of the class meetings on the schedule for the course Information Extraction 10-707 in Fall 2009.

Perceptrons as Margin Learners

• Slides for some of the talk.
• Notes for the rest.

Required Readings

• Large Margin Classification Using the Perceptron Algorithm, Freund and Schapire. No writeup required for this paper, since the project proposal is due today.

Optional Readings

• Exponentiated gradient algorithms for large-margin structured classification, Bartlett et al, NILS 2004. An alternative perceptron-like learning paradigm - multiplicate update, rather than additive update.
• A High-Performance Semi-Supervised Learning Method for Text Chunking, Ando and Zhang, ACL 2005. A fascinating use of "transfer learning" to improve NER performance.
• Robust Risk Minimization based Named Entity Recognition System, Zhang and Johnson, CoNLL 2003. One of several approaches to the CoNLL 03 NER task.
• A Simple Named Entity Extractor using AdaBoost, Carreras et al, CoNLL 2003. One of several approaches to the CoNLL 03 NER task.
• Named Entity Recognition using Hundreds of Thousands of Features, Mayfield et a, CoNLL 2003. One of several approaches to the CoNLL 03 NER task.