PBP: A Type of BP Neural Network Algorithm Based on Priori Triggering

Abstract:

The algorithm of anomaly detection for large scale networks is a key way to promptly detect the abnormal traffic flows. In this paper, priori triggered BP neural network algorithm(PBP) is analyzed for the purpose of dealing with the problems caused by typical algorithms that are not able to adapt and learn; detect with high precision; provide high level of correctness. PBP uses K-Means and PCA to trigger self-adapting and learning ability, and also, it uses historical neuron parameter to initialize the neural network, so that it use the trained network to detect the abnormal traffic flows. According to experiments, PBP can obtain a higher level of correctness of detection than priori algorithm, and it can adapt itself according to different network environments.

• Periodical: Advanced Materials Research (Volumes 271-273)
• Pages: 441-447
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.271-273.441
• Online since: July 2011
• Authors: Xiao Mei Chen, Dang Gang, Tian Yang
• Keywords: Network Traffic Anomaly Detection, Priori Algorithm, Self-Learning

References

[1] Ling Huang, Xian Long Nguyen, Minus Garofalakis, Jaseph Hellerstein. Communication-Efficient Online Detection of Network-Wide Anomalies. INFOCOM 2007. 26th IEEE International Conference on Computer Communications. (2007): 134-142.

DOI: 10.1109/infcom.2007.24

[2] Marina Thottan, Chuanyi Ji. Anomaly Detection in IP Networks. IEEE Transactions On Signal Processing. 2003, 51(8): 2191-2204.

DOI: 10.1109/tsp.2003.814797

[3] Tapas Kanungo, David M. Mount, Nathan S. Netanyahu, Christine D. Piatko, Ruth Silverman, Angela Y. Wu. An Efficient k-Means Clustering Algorithm: Analysis and Implementation IEEE Transactions on Pattern Analysis and Machine Intelligence. 2002, 20(7): 881-892.

DOI: 10.1109/tpami.2002.1017616

[4] Zheng Jun, Hu Mingzeng, Yun Xiaochun, Zheng zhong, Large Scale Network Traffic Anomaly Detection Based On Data Flow Methord , Transaction on Communications, 2006, 27(2): 1-8.

[5] Lakhina A, Crovella M, Diot C．Characterization of Network-Wide Anomalies in Traffic Flows, Technical Report BUCS-2004-020, Boston University, (2004).

DOI: 10.1145/1028788.1028813

[6] Lakhina A, Mark Crovella, Christophe Diot. Diagnosing network wide traffic anomalies Applications, Technologies, Architectures, and Protocols for Computer Communication Proceedings of the 2004 conference for computer communications. 2004: 219-230.

DOI: 10.1145/1015467.1015492

[7] Augustin Soule, Kav´e Salamatian, Nina Taft. Combining Filtering and Statistical Methods for Anomaly Detection. in IMC 05, Berkeley, Oct. 05.

DOI: 10.1145/1330107.1330147