A study on network fault knowledge acquisition based on support vector machine

Network fault knowledge acquisition is a necessary part of intelligent network management. In the paper, knowledge acquisition of two hierarchies is designed for modern network of large scale and some performance parameters instead of management information base are used to model the network faults so that the evaluations of network fault knowledge acquisition can easily be uniformed. Our knowledge acquisition methods are based on support vector machine. Basic support vector machine learning is applied to local network fault knowledge acquisition, and incremental PSVM is improved to be adapted to global dynamic network fault knowledge acquisition. Simulations indicate the correctness and efficiency of our method and the global network fault knowledge acquisition based on proximal support vector machine is still to be improved further.