Computer Network Intrusion Anomaly Detection Based on Rough Fourier Fast Algorithm

Since there is a close relationship between network information security attack events and time complexity, it necessary to count the degree of correlation current connection record within certain period before. Only in this way can data be better reflected. In paper, rough Fourier fast algorithm based on set theory proposed. Based characteristic attributes intrusion detection with most value character as division basis, computer anomaly intelligently divided into small sets, so carry out attribute reduction. The rule update experiment adopts misuse method, extracts some samples from KDD99 for training, obtains rules hierarchical decision table, uses incremental learning rules, compared expressed by table test feasibility effectiveness update, improved RSDB, RE-RFE algorithm, KNN evaluate effect model applied problem detection. Using subset reduced perform classification modeling significantly reduced, average 0.09 seconds, paper lays good foundation application algorithm.