Cyber security is the most often threat found in the real world environment. Some of the examples of cyber security assaults are credit card fraudulent activities, net banking foraging, firewall cracking and so on. These fraudulent activities can be detected by finding the patterns of theft activities. This can be attained by analysing and learning the transaction records collected from the users for different time period. There are many machine learning algorithms are introduced earlier for the prediction of cyber security threats. Support vector machine is the most popular machine learning algorithm utilized in various research works for the accurate prediction of patterns of cyber security threat. However SVM tends to have more computational overhead. Better outcome of SVM can be obtained by adjusting the configuration parameter values of SVM. This is focused and resolved in this research work by introducing the method namely Big Data Cyber Security Framework (BDCSF). In this research work credit card transaction is considered for the cyber security application. Initially credit card transaction information is gathered from the millions of users to ensure the security framework. This dataset is pre-processed using entropy based discretization procedure. After pre-processing artificial bee colony algorithm is utilized to perform the feature selection process. These selected features are then learnt using the SVM algorithm. The simulation assessment of the proposed and existing techniques tend to prove the proposed method seems to have better security performance than the existing method.
Volume 11 | Issue 10