1
$\begingroup$

How does one tell if a given model is generative AI or predictive AI?

Do generative models have more outputs than inputs and vice versa for predictive models?

$\endgroup$

1 Answer 1

2
$\begingroup$

It's hard nowadays to draw a distinct line between the two with the advent of conditional generation and normalizing flow

Usually, we say that is generative if you try to model the joint probability $p(x,y)$, whereas discriminative models try to model the marginal $p(y|x)$

However, generative model aims at sampling from such learned distribution, where usually instead discriminative models are only interested in some statistics of it (such as the mode)

$\endgroup$
2
  • $\begingroup$ I would lead with and emphasize sampling, a lot of sequence generators are predictive models of distribution, plus a separate sampling routine $\endgroup$ Commented Jun 19 at 7:29
  • 1
    $\begingroup$ @NeilSlater yup, though they start from the hypothesis of modelling $p(x)$ and they factor it as $\prod p(x_i | x_{j<i})$, and so it falls back to the usual classification problem $\endgroup$
    – Alberto
    Commented Jun 19 at 10:13

You must log in to answer this question.

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