This course will introduce students to the principles of statistical and syntactic pattern recognition. After a brief review of the principles of probability, random variables and vectors, we will study Bayes decision theory and criteria for classification. We will then consider the theory of maximum likelihood and Bayesian learning for parametric pattern recognition. After that, we will focus our attention to non-parametric methods such as classification using nearest neighbour rules and discriminant functions. The use of these in Neural Network Classifiers will be highlighted. The course will also introduce students to various features used in speech, shape and character recognition. With regard to syntactic pattern recognition we will briefly study the use of distance and probabilistic criteria in classifying strings, substrings, subsequences and trees as used in speech recognition and in matching RNA sequences. We will finally consider probabilistic classification of linear and syntactic patterns. The student shall conduct a project dealing with recent (up to within the last few months) applications of statistical and syntactic pattern recognition in security and communications.
After completing the course, the student is expected to: have advanced theoretical and practical skills in statistical and syntactic pattern recognition be able to solve research problems within security and communications using statistical- and syntactic pattern recognition based tools