A Review of Kernel Methods Based Approaches to Classification and Clustering of Sequential Patterns, Part II: Sequences of Discrete Symbols

A Review of Kernel Methods Based Approaches to Classification and Clustering of Sequential Patterns, Part II: Sequences of Discrete Symbols

Veena T. (Veena T.Indian Institute of Technology Madras, India), Dileep A. D. (Dileep A. D.Indian Institute of Technology Madras, India) and C. Chandra Sekhar (Indian Institute of Technology Madras, India)
DOI: 10.4018/978-1-61350-056-9.ch003
OnDemand PDF Download:
$37.50

Abstract

Pattern analysis tasks on sequences of discrete symbols are important for pattern discovery in bioinformatics, text analysis, speech processing, and handwritten character recognition. Discrete symbols may correspond to amino acids or nucleotides in biological sequence analysis, characters in text analysis, and codebook indices in processing of speech and handwritten character data. The main issues in kernel methods based approaches to pattern analysis tasks on discrete symbol sequences are related to defining a measure of similarity between sequences of discrete symbols, and handling the varying length nature of sequences. We present a review of methods to design dynamic kernels for sequences of discrete symbols. We then present a review of approaches to classification and clustering of sequences of discrete symbols using the dynamic kernel based methods.
Chapter Preview
Top

Design Of Dynamic Kernels For Discrete Symbol Sequences

The main issue in designing a kernel for discrete observation symbol sequence is to handle the varying length nature of the sequences. The varying length sequences of discrete observation symbols may be explicitly mapped onto a fixed dimensional feature vector and then the kernel is computed as an innerproduct in that fixed dimensional space. Instead of obtaining an explicit feature map, kernel between a pair of discrete symbol sequences can also be computed implicitly either by defining a function or an operation between the pair of sequences. In this section we present the design of various dynamic kernels for discrete symbol sequences.

Complete Chapter List

Search this Book:
Reset