Loading...

A PSO-OSELM based Machine Learning Method for Internet Traffic Classification

Al Shammari, Amir Abdollah | 2021

323 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 53813 (52)
  4. University: Sharif University of Technology, International Campus, Kish Island
  5. Department: Science and Engineering
  6. Advisor(s): Peyvandi, Hossein
  7. Abstract:
  8. Classification of Traffic Internet obtained early interest in the computer science community. Various methods have been presented for classifying the traffic of Internet to manage both security and Quality of Service (QoS). Nonetheless, traditional methods of classification including scheme of Transmission Control Protocol/Internet Protocol (TCP/IP) have not been accepted because of their complicated management. Classification method of network through learning algorithms of machine is the most popular classification method of traffic at this time. ELM was proposed as a modern algorithm of learning for the Single-hidden Layer Feed Forward Neural Networks (SLFNs). Meanwhile, learning process and SLFNs refresh loads of network according to gradient. Nevertheless, in ELM, sizes of input and biases are selected randomly, and sizes of output are computed with the analytical method contrary to SLFNs. In the current thesis, methods of ELM were utilized for Internet traffic classification. The Online Sequential OSELM, one of the approaches of ELM was employed. Particularly, Particle Swarm Optimization Algorithm (PSO) based software (PSO-OSELM) was extended for parameters selection used in (OSELM) algorithm. Finally, the proposed method is compared with another similar algorithm and the experimental results show the feasibility and effectiveness of the proposed method. It was seen that an accuracy rate of over 90.05% was achieved along with the application developed with PSO
  9. Keywords:
  10. Extreme Learning Machine ; Traffic Classification ; Particles Swarm Optimization (PSO) ; Online Sequential ; Service Quality

 Digital Object List

 Bookmark

No TOC