Linguistics
L700: Machine Learning Approaches in CL

In recent years, Machine Learning (ML) techniques have become very
influential in Computational Linguistics (CL). This course will
introduce some of the major supervised and unsupervised algorithms
as well as applications of these techniques to a wide range of CL
topics. The course will start with an introduction to ML and some
insights into complexity issues. The schedule of the course will be
decided in the first week of classes, students' interests
determining the final choice of applications to be covered in the
course. Possible supervised algorithms are: decision trees, k-
nearest neighbor learning, maximum entropy, memory-based learning,
Naive Bayes, neural networks, Ripper, transformation-based learning,
and Winnow. Possible unsupervised learning methods include
conceptual clustering, latent semantic clustering, and minimum
description length. Further topics that will be covered concern the
problem of evaluation and issues in data size and optimization.

Prerequisite: Some programming experience.

Optional Textbooks:

Tom Mitchell (1997) Machine Learning. McGraw-Hill.
Walter Daelemans and Antal van den Bosch (2006) Memory-Based
Language Processing. Cambridge University Press.