International Journal of Advanced Science and Technology Vol.83 (2015), pp.27-40 http://dx.doi.org/10.14257/ijast.2015.83.03 ISSN: 2005-4238 IJAST Copyright ⓒ 2015 SERSC Neural Network Based Intelligent Retrieval System for Verifying Dynamic Signatures Ankita Wadhawan and Avani Bhatia Assistant Professor, Department of Information Technology DAV Institute of Engineering & Technology, Jalandhar, Punjab, India ankitawadhawan6@yahoo.co.in,bhatia_avani@yahoo.com Abstract Mining of data is the process of discovering patterns from a large data set and uses this knowledge for matching purpose. In this paper a neural network based approach of data mining is used to verify dynamic signature patterns. In an authentication process, everyone may have a signature that is used to legally prove the document and to bind the individual with the inclination contained in the document. Signature verification is the verification process in which a given input is examined and is either rejected as forgery or accepted as genuine. The proposed algorithm is applied to a set of 500 signature samples collected from 20 individuals. Performance of the system is depicted by using three parameters that are accuracy, false acceptance rate (FAR) and false rejection rate (FRR). Experiments are performed by training the system with more and more number of samples. The results show that the system with neural network has better performance as compared to support vector machines. Keywords: Online signature verification, NN, MLP, FRR, FAR 1. Introduction Data mining is the process of discovering knowledge by extracting hidden patterns from large databases. It is applied in various machine learning applications like natural language processing, syntactic pattern recognition, fraud detection etc to predict information from past and present records. Neural networks are biological systems used to recognize patterns take decisions and learn and artificial neural networks are computer programs that implements machine learning algorithms to derive predictive models. The requirement for a genuine means of personal identification presents a challenge to almost every modern organization. Multimedia application developers and an engine for online signature verification are provided by this system. Online verification uses shape of an individual's signature and also logs the pen timing throughout the signing process. Characteristics of Biometric Features are:  Features are as unique as possible i.e., they don‟t match with any other person.  Occur in as many people as possible  Appropriate and easy to measure.  Measurable with simple technical instruments  They don‟t change with changes. Online signatures are written with an electronic device and the dynamic information is available at high resolutions. Data such as the direction captured dynamically, stroke, pressure, and shape of an individual‟s signature can enable handwriting to be a genuine indicator of one‟s identity. Advantages if dynamic signatures over static signatures: