Learning to upgrade internet information security and protection strategy in big data era

Abstract Nowadays, we are in a typical information age, and network and information security systems are facing severe challenges. However, there is currently no suitable method for the detailed analysis and protection of Internet information security. In order to analyze Internet security information more accurately, we proposed an improved KPCA (Kernel Principle Component Analysis) algorithm within the context of the big data era and in view of the shortcomings and disadvantages of the KPCA feature extraction algorithm. The proposed algorithm not only retains its performance ability, but also improves the subsequent classification ability. This paper uses the KDDCUP99 security audit data set to simulate network intrusions, and the data set used network information data resources within a total of 9 weeks. The training data set contains a total of 7 weeks of data information, and the other 2 weeks of data information are used as a validation data set. The training data set contains a total of 5 million records of network security information, while the verification data set contains 2 million records. The experimental results show that for the network intrusion classification test, the improved algorithm is more efficient, convenient, and faster than the traditional KPCA one. Furthermore, the simulation results also show that the proposed algorithm has achieved a very high degree of accuracy and improvement in terms of “accuracy rate,” “false alarm rate,” and “missing alarm rate.”