Intelligent Systems

Advances in Biometric Systems, Soft Computing, Image Processing, and Data Analytics

Chiranji Lal Chowdhary, Chiranji Lal Chowdhary

• Editorial: Apple Academic Press
• Año: 2019
• ISBN:
  9780429560040
• Edición: 1
• Categoría: Business
• Categoría: Gestione dell'informazione

This volume helps to fill the gap between data analytics, image processing, and soft computing practices. Soft computing methods are used to focus on data analytics and image processing to develop good intelligent systems. To this end, readers of this volume will find quality research that presents the current trends, advanced methods, and hybridized techniques relating to data analytics and intelligent systems. The book also features case studies related to medical diagnosis with the use of image processing and soft computing algorithms in particular models.

Providing extensive coverage of biometric systems, soft computing, image processing, artificial intelligence, and data analytics, the chapter authors discuss the latest research issues, present solutions to research problems, and look at comparative analysis with earlier results. Topics include some of the most important challenges and discoveries in intelligent systems today, such as computer vision concepts and image identification, data analysis and computational paradigms, deep learning techniques, face and speaker recognition systems, and more.

PART I

Biometric Systems and Image Processing

Chapter 1

Intelligent Techniques: An Overview

T. K. DAS^1* and D. K. BEBARTA²

¹ School of Information Technology and Engineering, VIT University, Vellore, India

² Department of Computer Science and Engineering, GVPCEW, Vishakhapatnam, India

* Corresponding author. E-mail: [email protected]

Abstract

With the impression of enhancing the proficiency and effectiveness of image processing and biometric system, this article provides an insight to the leading intelligent techniques that could be employed to enrich the biometric image analysis process. Leading techniques like expert system, artificial neural network, fuzzy system, genetic algorithm, and rough computing are being investigated in this review for their suitability of implementing in a biometric system. However, one technique does not fit for all problem domains, hence basing on the complexity involved in underlying data; it has been found as which technique could be employed for a particular problem to elicit the desired knowledge out of it. This study can be helpful for designing an intelligent system, which can be tailored for the organizations basing on their domain and nature of data.

1.1 Introduction

Conventional computing known as hard computing needs an accurate analytical model with a lot of computation time. Real world problems exist in a non-ideal environment fails to compute accurately by many analytical models. Soft computing (SC) differs from hard computing and unlike hard computing, it is tolerant of imprecision, uncertainty, partial truth, and an approximation. The guiding principle of SC is to exploit the tolerance for imprecision, uncertainty, partial truth, and approximation to achieve tractability, robustness and low computation cost. The names of several SC tools are fuzzy systems, neural networks, evolutionary computation, machine learning, and probabilistic reasoning. Different SC tools can be used in different phases of the planned analytical models. Information processing and analysis such as removing noise, hierarchical classification and clustering, searching, decision-making, and predicting the data in order to build a smarter, efficient, adaptable system to assist human in decision-making in the fields of information systems and business intelligence, internet computing, image processing, robotics, systems and control, bioengineering, and financial services and engineering. Different SC techniques can be used to solve above-mentioned issues such as wave-lets for removing noise, wavelets, fuzzy logic, and neural network for the hierarchical classification and clustering, evolutionary algorithms for searching, fuzzy systems for decision-making, artificial intelligence (AI) specifically suited to different tasks, such as waveform analysis, monitoring electronic data streams in the field of healthcare, energy market, currency exchange, stocks, and several other nonlinearity, unusual high volatility, and chaotic nature of data to predict the important trends.

SC techniques such as expert systems (ES), case-based reasoning (CBR), artificial neural networks, genetic algorithm (GA), fuzzy systems are considered as AI techniques since it involves some kind of human like learning, decision-making, and acting. These techniques assist decision makers to select effective actions in real-life especially in critical decision scenarios. Besides this, it reduces information overflow, facilitate current information; enable communication required for collaborative decisions; and deal with uncertainty in decision problems. Diverse range of intelligent techniques are represented in Figure 1.1.

FIGURE 1.1 Intelligent techniques.

Source: Das (2016).

1.2 Expert System

The inception of ES was long back in the year of 1972. The idea was to design a computer-based system which would work like a human expert. ES is a tool which augments human like decision-making by using AI techniques. It is programmed in such a way that it could be able to achieve the human logical thinking ability. It is a computer-based system programmed to use AI technique. Initially knowledge is acquired from domain experts and the acquired knowledge is represented in various forms such as rules, frames, or semantic nets. The heart of ES is inference engine which generates inference out of knowledgebase. An abstract view of ES is represented in Figure 1.2.

FIGURE 1.2 Abstract view of an expert system.

ES is being built for a variety of purposes, including medical diagnosis, mineral exploration, system fault finding, and many more. It is used in a context when a person requires some expertise to solve a problem. It is beneficial when human experts are expensive and difficult to find. However, it is quite useful in a situation of unpleasant environment and monotonous operation. One of the first generation of ES meant for chemical analysis; DENDRAL (Liang and Mahmud, 2012) Healthcare domain ES monitors surgeries, diagnosis the problem, and suggest actions. ES also automatically collect, aggregate, and analyze data for detailed analysis. ES, for example, MYCIN (Shortliffe, 1976), PROSPECTOR (Duda and Reboh, 1984) can handle uncertainty aspect of the underlying data as well.

CBR is quite effective in analyzing and designing decision support system for diagnosis in healthcare domain (Bichindaritz and Marling, 2006). CBR systems have been designed to look at the complexity of biomedicine, to directly integrate into clinical settings for closely monitoring and communicating with diverse systems. CBR system usefulness in decision-making in integrating with web 2.0 has been studied by He et al. (2009).

1.3 Artificial Neural Network

Artificial neural network known as ANN comprises a collection of simple nonlinear (or linear) processing components called neurons, which are linked together via a net of adaptable connections. A neuron can transform a set of input to a reasonable output depending on the properties of activation function it uses. A set of neurons act as input nodes and another set act as output nodes. In between input and output layers, the nodes process information by weighted summation, and thresholding. These operations are guided by the principle of a leaning algorithm basing on which the network functions. Learning for a network is known as training. Like human being, the network is being trained in order to accomplish a particular task. A multilayer neural network is shown in Figure 1.3.

FIGURE 1.3 The ...

Índice

• Cover
• Half Title
• Title Page
• Copyright Page
• Dedication
• About the Editor
• Table of Contents
• Contributors
• Abbreviations
• Preface
• Acknowledgments
• PART I: Biometric Systems and Image Processing
• PART II: Soft Computing and Data Analytics
• PART III: Intelligent Systems and Hybrid Systems
• Index