A mobile application identification method based on adaptive incremental learning

A mobile application and incremental learning technology, applied in the direction of neural learning methods, biological neural network models, etc., can solve the problems of difficult classifiers and difficult control of classifier parameters, so as to avoid retraining process and catastrophic forgetting , the effect of improving reliability

Inactive Publication Date: 2019-05-07
GUANGZHOU NASSOFT INFORMATION TECH CO LTD
View PDF0 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

One solution is to add the traffic of the new mobile application to the original training set when a new mobile application type appears, and then retrain a deep learning classifier, but the parameters of the new classifier become difficult to control , it will become extremely difficult to train a qualified new classifier

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A mobile application identification method based on adaptive incremental learning
  • A mobile application identification method based on adaptive incremental learning
  • A mobile application identification method based on adaptive incremental learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0020] The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments, where the schematic embodiments and descriptions of the present invention are used to explain the present invention, but not to limit the present invention.

[0021] A mobile application recognition method based on adaptive incremental learning, including the following five steps:

[0022] Step 1: Train a basic deep model that can complete mobile application type recognition, and at least one binary classification problem.

[0023] Step 2: When the deep model obtained in step 1 performs classification tasks, there may be data identified as "unknown application type", and the system will retain these data, according to the triplet (source IP address, destination IP address, destination port) classification, data in the same class will be considered to come from the same application.

[0024] Step 3: The data from the same mobile application obt...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a mobile application identification method based on adaptive incremental learning. The method is a method for expanding the identification range under the condition that the number of to-be-identified mobile applications is increased based on the incremental learning (Incental Learning). The objective of the invention is to solve the catastrophic forgetting phenomenon of the existing mobile application identification technology, when a new mobile application type identification task is given, an original deep learning classifier with a good effect is kept in an originalstate, and a new mobile application type is added step by step on the basis of the original deep learning classifier. Therefore, the whole mobile application identification system has the self-adaptive incremental learning capability and the capability of expanding the number of the types of the to-be-identified mobile applications.

Description

technical field [0001] The invention relates to a mobile application identification method based on self-adaptive incremental learning. Background technique [0002] The current research on mobile application recognition based on deep learning methods extracts and classifies data features at the data packet level, and uses a stacked autoencoder with a relatively good sequence feature extraction effect to perform feature extraction on data packets, such as figure 1 shown. Because after the automatic encoder is learned, the output part is approximately equal to the input in performance, so that the traffic characteristics learned by the hidden layer are closer to the nature of the traffic data. Since the hidden layer can express the characteristics of the data traffic, the data traffic can be classified into corresponding mobile application categories through the hidden layer. For 15 different mobile applications with a total of 12GB of traffic as the training set, a deep le...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06N3/08
Inventor 刘宁李大燊闫高峰高巍
Owner GUANGZHOU NASSOFT INFORMATION TECH CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap