I know you search a lot on the internet about these topics, and most of them are very confusing and complexing for newbies. I am briefly explaining these hot topics so; you can understand the basic working and difference between both.
GAN’s are the most progressive method of neural networks. Basically, GANs consist of two main neural networks one is called the generator, and the other called discriminator. In recent GAN’s networks are mainly used for generating synthetic information or fake data. The generator’s main objective is to generate fake data from noise data. It will generate images or text or videos.
Whereas the discriminator’s purpose is to find the input data or information is fake or real, it is like a binary classifier. You can say that it can act as a binary CNN's network, finally, it will classify whether the input data or info is real or not.
The second figure is basically the reverse of the first figure where the data was generated through the noise. Now the following picture demonstrates the full working principle of GANs.
Now, let’s come to CNN's; CNN's is basically a single model contains three layers. The first layer is input, the second layer is called hidden layers, and the last layer called the output layer. Remember one thing here, the hidden layer could be a set of multiple layers where each layer can contain multiple neurons. CNN's are designed for data with spatial structure. For example, images, which have a natural spatial ordering to it are perfect for this network. The following figure is to make understand more about CNNs.
CNN's is the one part of the GANs network, whereas GANs having both the networks CNN ad DCNN. Hope you understand well about the basics of these.
إرسال تعليق