What is the Discriminator in a GAN modelling? I mean, what pattern is the Discriminator network finding that the Generator network won't find? The Generator network is already learning what real images look like and don't look like, no? Or is the Gen only learning what is is but not what isn't? Also, is the Discriminator modelling pure Accuracy? It seems to have a binary classification and is said to model "less", like as if it's not storing pattern features but rather just attaching tags to the Gen net of what is most probable.......?? Is a GAN's Discriminator network used for identifying key things in an image that prove its Fake? Like a cat with a beard? Is this ALL it does? Really helps me if so....
A simple markov chain / tree trie / PPM models frequency of strings, and another pattern which is less common is family last names, so that's why I'm asking what exactly is a GAN finding? What is it mixing in to the soup? ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Td13181dbab529095-M40c2583aa03b9b87c46646fb Delivery options: https://agi.topicbox.com/groups/agi/subscription
