Network Forensic Computing Based on ANN-PCA

The network forensic computing is faced with the question of the massive information stream analyses. Two solutions including feature extracting and classification mining were proposed. The main work includes features extracting with ANN-PCA, classification mining after features extracting in FAAR algorithm which we proposed to mine association rules based on PCA and fuzzy classification technology. The experiment indicates that the classification accuracy raise distinctly after feature extracting, and the feature quantity decrease especially U2R attack. Further more, the number of generating rules also reduces clearly.