I'm trying to model a generative model for classification problem, especially aiming to solve an imbalanced data problem.

However, I couldn't get intuitive understanding for generative classifier in terms of deep learning (or machine learning). Classical generative classifiers such as Bayes classifier are clear with such a closed form of equations.

In a viewpoint of 'neural network', it is necessary to model a network structure with input, output, and loss function to update parameters. Assuming class distribution $p(y)$ is known, network should be like below, I think:

enter image description here

Q1) I wonder that the network structure makes sense for classification even though the diagram is very simplified version. Additionally, how can I return the log likelihood function $\log p_{\theta} (x | y)$ as an output layer in keras? (How can I constitute the network structure in keras?)

Q2) How is loss function defined to update $\theta$ ?

  • $\begingroup$ You can use latex on this site. Please, edit your post to do that. $\endgroup$ – nbro Dec 19 '20 at 12:49
  • 1
    $\begingroup$ @nbro I just edited to use latex. Thank you $\endgroup$ – WKIm Dec 20 '20 at 22:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.