GANs consist of three terms Generative Adversarial Network. Let's understand these three terms first. Generative : A Generative Model takes input training sample from some distribution and learns to represents that distribution. Adversarial : It basically means Conflicting or Opposing. Networks : These are basically neural networks. So,Generative Adversarial Networks are deep neural network architecture comprising of two neural networks compete with each other to make a generative model. A GAN consist of two class models : Discriminative Model :- It is the one that discriminate between two different classes of data.It tries to identify real data from fakes created by the generator Generative Model :- The Generator turns noise into an imitation of the data to try to trick the discriminator Mathematically, A Generative Model 'G' to be trained on training data 'X' sampled from some true distribution 'D' is the one which, given some standard random distrib...
GANs consist of three terms Generative Adversarial Network. Let's understand these three terms first.
Generative : A Generative Model takes input training sample from some distribution and learns to represents that distribution.
Adversarial : It basically means Conflicting or Opposing.
Networks : These are basically neural networks.
So,Generative Adversarial Networks are deep neural network architecture comprising of two neural networks compete with each other to make a generative model.
A GAN consist of two class models :
- Discriminative Model :- It is the one that discriminate between two different classes of data.It tries to identify real data from fakes created by the generator
- Generative Model :- The Generator turns noise into an imitation of the data to try to trick the discriminator
Mathematically,
A Generative Model 'G' to be trained on training data 'X' sampled from some true distribution 'D' is the one which, given some standard random distribution 'z' produces a distribution 'd' which is close to 'D' according to some closeness metric.
Intuition Behind GANs
- Generator starts from noise to try to create an imitation of the data.
- Discriminator looks at both real data and fake data created by the Generator
- Discriminator tries to predict what's real and what's fake.
- Generator tries to improve its imitation of the data using Back propogation
- Discriminator tries to identify real data from fakes created by the generator
- After training Generator network is used to create new data that's never been seen before.
Comments
Post a Comment