ABSTRACTCurrent neural network technology is the most progressive of the artificialintelligence systems today. Applications of neural networks have made thetransition from laboratory curiosities to large, successful commercialapplications. To enhance the security of automated financial transactions,current technologies in both speech recognition and handwriting recognition arelikely ready for mass integration into financial institutions. RESEARCH PROJECTTABLE OF CONTENTS Introduction 1 Purpose 1 Source of Information 1 Authorization 1 Overview 2 The First Steps 3 Computer-Synthesized Senses 4 Visual Recognition 4 Current Research 5 Computer-Aided Voice Recognition 6 Current Applications 7 Optical Character Recognition 8 Conclusion 9 Recommendations 10 Bibiography 11INTRODUCTIONPurposeThe purpose of this study is to determine additional areas where artificialintelligence technology may be applied for positive identifications ofindividuals during financial transactions, such as automated bankingtransactions, telephone transactions , and home banking activities. This studyfocuses on academic research in neural network technology . This study wasfunded by the Banking Commission in its effort to deter fraud.Order now
OverviewRecently, the thrust of studies into practical applications for artificialintelligence have focused on exploiting the expectations of both expert systemsand neural network computers. In the Artificial Intelligence Essay community, theproponents of expert systems have approached the challenge of simulatingintelligence differently than their counterpart proponents of neural networks. Expert systems contain the coded knowledge of a human expert in a field; thisknowledge takes the form of “if-then” rules. The problem with this approach isthat people don’t always know why they do what they do. And even when they canexpress this knowledge, it is not easily translated into usable computer code. Also, expert systems are usually bound by a rigid set of inflexible rules whichdo not change with experience gained by trail and error.
In contrast, neuralnetworks are designed around the structure of a biological model of the brain. Neural networks are composed of simple components called “neurons” each havingsimple tasks, and simultaneously communicating with each other by complexinterconnections. As Herb Brody states, “Neural networks do not require anexplicit set of rules. The network – rather like a child – makes up its ownrules that match the data it receives to the result it’s told is correct” (42). Impossible to achieve in expert systems, this ability to learn by example is thecharacteristic of neural networks that makes them best suited to simulate humanbehavior.
Computer scientists have exploited this system characteristic toachieve breakthroughs in computer vision, speech recognition, and opticalcharacter recognition. Figure 1 illustrates the knowledge structures of neuralnetworks as compared to expert systems and standard computer programs. Neuralnetworks restructure their knowledge base at each step in the learning process. This paper focuses on neural network technologies which have the potential toincrease security for financial transactions. Much of the technology iscurrently in the research phase and has yet to produce a commercially availableproduct, such as visual recognition applications. Other applications are amultimillion dollar industry and the products are well known, like SprintTelephone’s voice activated telephone calling system.
In the Sprint system theneural network positively recognizes the caller’s voice, thereby authorizingactivation of his calling account. The First StepsThe study of the brain was once limited to the study of living tissue. Anyattempts at an electronic simulation were brushed aside by the neurobiologistcommunity as abstract conceptions that bore little relationship to reality. This was partially due to the over-excitement in the 1950’s and 1960’s fornetworks that could recognize some patterns, but were limited in their learningabilities because of hardware limitations.
In the 1990’s computer simulations ofbrain functions are gaining respect as the simulations increase their abilitiesto predict the behavior of the nervous system. This respect is illustrated bythe fact that many neurobiologists are increasingly moving toward neural networktype simulations. One such neurobiologist, Sejnowski, introduced a three-layernet which has made some excellent predictions about how biological systemsbehave. Figure 2 illustrates this network consisting of three layers, in whicha middle layer of units connects the input and output layers.
When the networkis given an input, it sends signals through the middle layer which checks forcorrect output. An algorithm used in the middle layer reduces errors bystrengthening or weakening connections in the network. This system, in whichthe system learns to adapt to the changing conditions, is called back-propagation. The value of Sejnowski’s network is illustrated by an experiment byRichard Andersen at the Massachusetts Institute of Technology. Andersen’s teamspent years researching the neurons monkeys use to .