Research On Abnormal Detection Of Network Finance Based On Big Data

In recent years,with the rapid development of Internet technology,especially the rapid development of mobile Internet has brought a higher level of information age.With the network-based financial transactions such as electronic payment,mobile payment development,it also brought a more severe financial transaction security problem.It is necessary to study the anomaly detection of network financial transactions based on large data,and to study the application of anomaly detection algorithm in the field of network financial transactions to accurately identify the network financial transactions,and prevent the financial losses caused by illegal financial behavior.Up to now,although many kinds of anomaly detection algorithms have been proposed,there are still many problems in the practical application of network financial transactions,such as high computational complexity,low accuracy of the results and stability of the algorithm.In the case of large-scale network financial transaction anomaly detection,the existing model-based anomaly detection technology,such as distance-based or density-based proximity anomaly detection technology,there are still shortcomings in transaction data dimension disaster problem,data sparseness and massive datasets.So this paper is based on these aspects to carry out research.In order to solve the disaster of high-dimensional space dimension problem,we present an improved method for angle-based anomaly detection algorithm for large amount of data network financial transactions.Through the reduction of the edge of the angle for the calculation of angle variance,this method effectively reduces the computational cost,accelerate the speed of the algorithm,and decrease the requirements on time and physical memory.The current outlier mining approaches based on the distance or the nearest neighbor result in too long operation time results when using for the high-dimensional and massive data.So,we present a new anomaly detection algorithm based on the local distance of density-based sampling data.The existing local distance anomaly detection algorithm can be optimized.The probabilistic sampling of the required data set is carried out by using the probability sampling method based on density bias.Then,the local anomaly detection based on local distance is used to achieve the purpose of algorithm efficiency optimization.