Adaptive Network Structures for Data/Text Pattern Recognition (Application)

Adaptive Network Structures for Data/Text Pattern Recognition (Application)

Emmanuel Buabin (Methodist University College Ghana, Ghana)
DOI: 10.4018/978-1-4666-2661-4.ch023


The objective of this chapter is implementation of neural based solutions in real world context. In particular, a step-wise approach to constructing, training, validating, and testing of selected feed-forward (Multi-Layer Perceptron, Radial Basis function) and recurrent (Recurrent Neural Networks) neural based classification systems are demonstrated. The pre-processing techniques adopted in extracting information from selected datasets are also discussed. In terms of future practical directions, a catalogue of intelligent systems across selected disciplines, are outlined. The main contribution of this book chapter is to provide basic introductory text with less mathematical rigor for the benefit of students, tutors, lecturers, researchers, and/or professionals who wish to delve into foundational (practical) representations of bio-intelligent intelligent systems.
Chapter Preview


A data set (or dataset) is a collection of data, processed in tabular or non-tabular formats for scientific research purposes. In tabular datasets, each row is an instance whereas each column is an attribute. A Cell value is known as datum. Cell values may be labels, numeric or a combination of labels and numeric values. For supervised learning purposes, each row (i.e. instance) has a target. The target could be single or multiple. Unlike tabular forms, non-tabular datasets exist more in real life situations as text based datasets. Non tabular datasets are usually expressed in XML files for easy retrieval and processing. The University of California Irvine (UCI) Data repository makes available various datasets for research purposes, some of which have been explained in this book chapter.

Complete Chapter List

Search this Book: