INTOCS: Neural Network Inspired Oversampling based Secure and Optimized Transaction Framework for Credit Card

Online transactions have transformed commerce and international trade but have brought up new difficulties, especially regarding fraud detection and security. The exponential rise in digital transactions has made reliable fraud detection systems using machine learning and artificial intelligence techniques which are leveraged to detect and mitigate fraud thereby increasing the security of transactions. The paper suggested a model in which data preprocessing using standard scalar along with SMOTE for data balancing is proposed. Data is split into 80-20% train test data based on credit card transactions. An artificial neural network is used to determine credit card fraud. Activation function followed by batch normalization blended with Adam Optimizer and cross-entropy is used to process the computation. Furthermore, classification parameters obtained an Accuracy of 99.95%, Precision of 99%, Recall of 99%, and F1Score of 99%. The result obtained by the proposed method outperforms the existing approach that focuses on secured transactions and helps both the client and server.