Gdfgdf

Journal of Theoretical and Applied Information Technology
© 2005 - 2009 JATIT. All rights reserved. www.jatit.org

BI DIRECTIONAL ASSOCIATIVE MEMORY NEURAL NETWORK METHOD IN THE CHARACTER RECOGNITION
G.Muni Sekhar Reddy Author: G.Muni Sekhar Reddy Yash Pal Singh$, V.S.Yadav$, Amit Gupta#, Abhilash Khare$ Email:munisekhar222@gmail.com

$ Bundelkhand Institute Of Engineering and Technology,Jhansi, INDIA #Krishna Institute of Engineering and Technology, Ghaziabad, INDIA Email: yash_biet@yahoo.co.in

ABSTRACT Pattern recognition techniques are associated a symbolic identity with the image of the pattern. In this work we will analyze different neural network methods in pattern recognition. This problem of replication of patterns by machines (computers) involves the machine printedpatterns. The pattern recognition is better known as optical pattern recognition. Since, it deals with recognition of optically processed patterns rather then magnetically processed ones. A neural network is a processing device, whose design was inspired by the design and functioning of human brain and their components. There is no idle memory containing data and programmed, but each neuron isprogrammed and continuously active. Neural network has many applications. The most likely applications for the neural networks are (1) Classification (2) Association and (3) Reasoning. One of the applications of neural networks is in the field of pattern recognition. The Bidirectional associative memory does heteroassociative processing in which, association between pattern pairs is stored. TheBidirectional Associative has capacity limitations. It can store and correctly recognize only six characters, with the condition that the characters should be slightly similar in shape. Keywords: Neural networks, bi-directional, machine printed patterns, pattern recognition 1. INTRODUCTION Statistical pattern recognition: Here, the problem is posed as one of composite hypothesis testing, each hypothesispertaining to the premise, of the datum having originated from a particular class; or as one of regression from the space of measurements to the space of classes. The statistical methods for solving the same involve the computation other class conditional probability densities, which remains the main hurdle in this approach. The statistical approach is one of the oldest, and still widely used [2].Syntactic pattern recognition: In syntactic pattern recognition, each pattern is assumed to be composed of sub-pattern or primitives strung together in accordance with the generation rules of a grammar characteristic of the associated class. Class identifications accomplished by way of parsing operations using automata corresponding to the various grammars [11, 12]. Parser design and grammaticalinference are two difficult issues associated with this approach to PR and are responsible for its somewhat limited applicability. Knowledge-based pattern recognition: This approach to PR [13] is evolved from advances in 382

A neural network is a processing device, whose design was inspired by the design and functioning of human brain and their components. There is no idle memory containing dataand programmed, but each neuron is programmed and continuously active. Neural network has many applications. The most likely applications for the neural networks are (1) Classification (2) Association and (3) Reasoning. One of the applications of neural networks is in the field of pattern recognition. Pattern recognition is a branch of artificial intelligence concerned with the classification ordescription of observations. Its aim is to classify patterns based on either a priori knowledge or on the features extracted from the patterns. Pattern recognition is the recognition or separation of one particular sequence of bits or pattern from other such patterns. Pattern recognition [PR] applications have been varied, and so also the associated data structures and processing paradigms. In...