AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
The Differences Between Continuous and Discontinuous Latent Variables
I like thinking about LDA in the topic modeling setting, because it's really illustrative of what we want to do when we're doing photographic modeling. Do you have any opinion on like discrete latent variables versus continuous latent variables? Maybe this is like a different setting, but it seems like with discrete latent variables, it might be more difficult to train a model than with continuous latent variables. But discrete ones might be more interpretable. Right.