Transmission Quality Classification with Use of Fusion of Neural Network and Genetic Algorithm in Pay&Require...

Area of Science:

• Computer Science
• Network Engineering
• Machine Learning

Background:

• Computer networks are essential for modern systems, with evolving transmission concepts like programmable networks.
• Assessing data transmission quality relies on technical parameters (delay, bandwidth, packet loss, jitter), which are complex for non-experts.
• Translating these technical metrics into user-understandable quality assessments is a significant challenge.

Purpose of the Study:

• To develop a Machine Learning (ML) model for dynamic classification of network data transmission quality parameters.
• To simplify the understanding of complex transmission metrics for non-expert users.
• To improve upon existing methods for assessing network service quality.

Main Methods:

• Utilized a hybrid approach combining a neural network (NN) with a genetic algorithm (GA) for weight selection, replacing traditional gradient descent.
• Trained the model on 100 samples, each with four features representing transmission parameters and a quality label.
• Employed 10-fold stratified cross-validation to evaluate classification performance.

Main Results:

• Achieved a high classification accuracy (SEN) of 95% for network data transmission quality.
• Reduced incorrect classifications to 5 out of 100 samples.
• Demonstrated superior performance compared to previous studies using single classifiers and ensemble learning.

Conclusions:

• The proposed ML model, integrating NN and GA, effectively translates complex network transmission parameters into understandable quality classifications.
• This approach significantly enhances user comprehension of network service quality.
• The achieved 95% accuracy indicates a robust and improved method for network quality assessment.