This course deals with pattern recognition which has several important
applications. For example, multimedia document recognition (MDR) and automatic medical
diagnosis are two such.
The emphasis of the course is on algorithms for pattern recognition.
The representation of patterns and classes and the proximity measures are an important aspect
of pattern recognition and are described in the earlier lessons.
When the data sets are very
large it is meaningful to reduce the data and used this reduced data for pattern classification.
The details of feature extraction and feature selection and prototype selection have been
In pattern recognition, we deal with classification and clustering of patterns. The
two well-known paradigms of machine learning namely, learning from examples or supervised
learning and learning from observations or clustering are dealt with in this course.
learning the classifiers such as nearest neighbour classifier, bayes classifier, decision trees and
support vector machines have been dealt with.
Clustering is an important aspect of unsupervised learning and has been
covered extensively in this course.Combination of classifiers have been
dealt with where more than one classifier is used to arrive at a class label.
The applications of
pattern recognition to a practical problem has been handled where the various techniques used on
a document recognition problem have been discussed.
Introduction – Definitions, data sets for Pattern
Different Paradigms of Pattern Recognition
Representations of Patterns and Classes
Metric and non-metric proximity measures
Different approaches to Feature Selection
Nearest Neighbour Classifier and variants
Efficient algorithms for nearest neighbour