PDF Publication Title:
Text from PDF Page: 045
4.1 Combination of convolutional and recurrent neural networks The novel applications of deep learning proposed in this thesis, for QoE and NTC prediction, are based on an architecture formed by combining CNN and RNN networks. This architecture combines two interesting properties that have demonstrated to be important to improve the classification capabilities of the resulting classifiers: • • The initial CNN layers are able to perform automatic knowledge representation. CNNs were initially applied to image processing, as a biologically inspired model to perform image classification, where feature engineering was done automatically by the network thanks to the action of a kernel (filter) which extracts location invariant patterns from the image. Chaining several CNNs allows extracting complex features automatically. A subsequent block formed by RNN layers allows to capture the time dependent information contained in the processed data. The RNN layers are specially indicated to process sequential information and have been applied mainly to textual and time-series data. The principles for the application of this combined architecture to network flow packets, are: - To transform the flow of data packets to an image-like data structure that can be processed by a CNN. We consider the matrix formed by the time-series of feature vectors as an image. Image pixels are locally correlated; similarly, feature vectors associated with consecutive time slots present a correlated local behavior, which allows us to adopt this analogy. - The data structure generated by the CNN block must be transitioned to a new data structure for the LSTM block. In this transition, it is important that the time dimension of the tensor produced by the CNN block is maintained and transformed to the time dimension of the input tensor of the LSTM block. This transition is performed by reshaping and permuting the dimensions of the tensors. - To perform a fine adjustment of the CNN and LSTM layers and their parameters. Adding more layers do not produce necessarily an improvement in classification performance. Fig 5 presents the combined CNN-RNN architecture. Of the different options available for the RNN block (LSTM, GRU...), we have chosen the LSTM model because its slight additional complexity (in comparison, for example, with GRU) is compensated by its better results. Doctoral Thesis: Novel applications of Machine Learning to NTAP - 43PDF Image | Novel applications of Machine Learning to Network Traffic Analysis
PDF Search Title:
Novel applications of Machine Learning to Network Traffic AnalysisOriginal File Name Searched:
456453_1175348.pdfDIY PDF Search: Google It | Yahoo | Bing
Cruise Ship Reviews | Luxury Resort | Jet | Yacht | and Travel Tech More Info
Cruising Review Topics and Articles More Info
Software based on Filemaker for the travel industry More Info
The Burgenstock Resort: Reviews on CruisingReview website... More Info
Resort Reviews: World Class resorts... More Info
The Riffelalp Resort: Reviews on CruisingReview website... More Info
CONTACT TEL: 608-238-6001 Email: greg@cruisingreview.com (Standard Web Page)