Mode Collapse - G vs D

Course Q&A Generative Adversarial Networks (GANS) Build Basic Generative Adversarial Networks
1

The Whole point of Generator is to generate the images that look as real as the realistic ones. if the discriminator gets fooled by the generator, the generator can create images of the similar kind or else has to change in the way it is generating images.
In the Mode Collapse lecture video,

What do we mean by this statement?
Take a discriminator that has learned to be good at identifying which handwritten digits are fakes, except for cases where the generated images look like ones and sevens.
From what I understood,So the discriminator at this iteration, it is classifying 1’s and 7’s as real even though they are fake.
Correct me If it doesn’t make sense

Thanks.

2

I have a bit of hardship understanding your question but in mode collapse means that the generator will more and more generate digits that are passing as real from the discriminator, so it will concentrate in creating those digit and leave others out because by doing so it has a has a guarantee for fooling the discriminator.

3

Is discriminator misclassification means if the discriminator classifies the image generated by the generator as real, then it is said to be misclassification?
Is this right?

4

Misclassification from discriminator if image was generated from generator and passed as real from discriminator, yeah you can say that.

5

Clearly understood, Thanks!