This study presents a predictive model for cloud computing security in the banking sector using Levenberg Marquardt Back Propagation algorithm trained with cuckoo search for fast and improved convergence speed. Object-oriented design methodology was used. The Levenberg Marquardt Back Propagation has been used to determine the training performance of an ANN, which is evaluated by computing the means square error of the system and that was the mean of the square of the difference between the target matrix and the input matrix. Cuckoo Search has been used to determine the weights of Neural Network. The signature feature vectors are input to the ANN; these features extracted from signature image were obtained via image processing. System was implemented in Matlab. Signature verification system based on the trained network was developed and tested with 160 signatures which consist of 90 genuine signatures,50 forgery signatures and 20 irregular signatures. The performance has been evaluated with False Rejection Rate of 0% and False Acceptance Rate of 8%.
展开▼
