Liu Chong

TL;DR: This work modeled the time-series network traffic by the recurrent neural network (RNN) and introduced the attention mechanism for assisting network traffic classification in the form of the following two models, the attention aided long short term memory (LSTM) as well as the hierarchical attention network (HAN).

TL;DR: Three weight calculation algorithms to explore the best feature weight update method based on synonym merging named SM-CHI are presented and it is proved the classification accuracy is the highest when choosing the improved CHI formula 2.

TL;DR: In this article, a virtual network mapping method and device based on an improved genetic algorithm was proposed. And the method comprises: encoding each virtual network node mapping scheme into a chromosome to obtain a plurality of encoded chromosomes; subjected each of the plurality of chromosomes to chiasma, variation, feasibility check and selection operation.

TL;DR: In this article, a network traffic classification method, a device and an implementation device are provided, which comprises the following steps: acquiring a data stream to be classified; the data stream includes at least one data packet; processing the stream to obtain a standard data stream; according to the traffic classification model established in advance, the standard data streams are classified.

TL;DR: In this article, the authors proposed a feature word vector obtaining method and a text classification method and device, which consists of calculating similarities between feature words in pairs in the feature word set, and dividing the feature words from the pair according to the similarities to obtain a plurality of synonym sets; aiming at each synonym set, calculating a target weight, aiming at a target text, of the synonyms set according to a first weight, and combining the target weight with a second weight, wherein the second weight is a weight, aimed at the target text of feature words not divided