Transaction fraud has grown to be an important issue in worldwide, banking and commerce security is easier access to trade information. Every day, there are more and more incidents of transaction fraud, which causes large financial losses for both consumers and financial professionals. The ability to identify transaction fraud is getting closer to reality due to improvements in computer science's machine learning (ML) and data mining areas. So, one of them that is becoming dangerous is credit card fraud (CCF). Millions of people are experiencing financial loss and identity theft as a result of these malicious operations. The CCF of many illegal activities that fraudsters are always using new methods to carry out. One major problem facing financial services sector is CCF. To overcome this, categorical boosting (CatBoost) algorithm is explained as a solution to these problems. Fraud or fraudulent transactions are identified using this effective CatBoost algorithm implementation for identification of CCF. Thus, in terms of accuracy, precision, and detection rate this method gives better performance.
Nowadays, intrusion detection systems (IDSs) have growingly come to be considered as an important method owing to their possible to expand into a key factor, which is crucial for the security of wireless networks. In wireless network, when there is a thousand times more traffic, the effectiveness of normal IDS to identify hostile network intrusions is decreased by an average factor. This is because of the exponential growth in network traffic. This is due to the decreased number of possibilities to discover the intrusions. This is because there are fewer opportunities to see possible risks. We intend an extreme learning machine with deep neural network (DNN) algorithm-based intrusion detection in clustering (EIDC) wireless network. The main objective of this article is to detect the intrusion efficiently and minimize the false alarm rate. This mechanism utilizes the extreme learning machine (ELM) with a deep neural network algorithm for optimizing the weights of input and hidden node biases to deduce the network output weights. Simulation outcomes illustrate that the EIDC mechanism not only assures a better accuracy for detection, considerably minimizes an intrusion detection time, and shortens the false alarm rate.