Search

An Easily-implemented Adaptive Packet Sampling (EAPS) is presented in this paper, which overcomes the shortcoming of NetFlow and Adaptive Netflow. EAPS is easy to be hardware-implemented and scalable for high-speed networks. Additionally, EAPS is automatically adaptive to traffic rate under resource constrains, thus it is convenient to be used by network operators. Experiments are conducted with the real network traces. Results show that EAPS is more accurate than ANF over links with light load or traffic fluctuations. [ABSTRACT FROM AUTHOR]

Under the guide of the novel biomimetics pattern recognition theory that is based on the character of human's recognition, combining the character of traditional neural network, a new multi-weight neural network is constructed to realize the idea of that theory. Extraordinary results are obtained with the first use of the new multi-weight neural network in speech recognition of finite words. The experiment results show that the multi-weight neural network can not only recognize finite words correctly and promptly, but also keep quite high correct recognition rate under the circumstance of small-number samples. [ABSTRACT FROM AUTHOR]