0
$\begingroup$

I had a task to implement a neural network that would carry out multiclass classification of traffic by several parameters. On the advice of colleagues, I chose the "Multilayer Perceptron" architecture. One of these days I will have a defense of my work, but I absolutely do not understand how to answer the question: "Why did you choose this type of architecture?". Please tell me if there are any theses why the "multilayer perceptron" architecture is better than other neural network architectures for solving problems of multi-class traffic classification?

$\endgroup$
2
  • $\begingroup$ That's hard to answer without more details about the problem. And when you say MLP, do you mean a network with threshold activation functions, or just a "vanilla" fully connected network? $\endgroup$
    – Lee Reeves
    Commented Jun 11, 2022 at 0:35
  • $\begingroup$ Hello! Thank you for your answer. The activation function is selected by ReLU. The neural network has the following structure: several neurons of input values, 2 fully connected layers of neurons and several resulting outputs for multiclass classification. It's just interesting to find out what is the advantage of MLP over other architectures in the tasks of multiclass traffic classification $\endgroup$
    – dremming
    Commented Jun 11, 2022 at 10:26

1 Answer 1

0
$\begingroup$

This is a very general question, so I'll just point to a reference that should be a good starting point. Deep Learning for Encrypted Traffic Classification: An Overview seems to contain exactly what you're looking for:

Several factors affect the choice of deep learning models for network traffic classification. The most important one is the choice of features. ...

Table II summarizes features, the corresponding models, and their properties.

$\endgroup$
1
  • $\begingroup$ Thank you so much! I will read this article. $\endgroup$
    – dremming
    Commented Jun 11, 2022 at 11:14

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .