What is deepfake technology? How it works - Disadvantages of using Deepfake Technology

What is deepfake technology? How it works - Disadvantages of using Deepfake Technology

Deepfake technology is one of the most popular application, which has a magical application, but did you ever think why Deepfake is getting the popularity?

Let me tell you, why Deepfake is getting that popularity and what is Deepfake technology? So Deepfake is basically part of deep-learning where you will actually do some of the amazing tasks and that can make a human fool as well.

So, what is Deepfake?

Deepfake technology is Synthetic Media in which, a person in an existing image or video is replaced with someone else's likeness. So if you want to change someone's face with like someone others face you can do that by using Deepfake. While the act of faking content is not new, Deepfakes leverage powerful techniques from machine learning and artificial intelligence to manipulate or generate visual and audio content with the high potential to deceive.

So here, what is Deepfake stands for?

Deepfake will become Deep Learning plus Fake if we break the name. So basically it's a part of artificial intelligence and we use the deep learning techniques like GAN to do all these things.

Now let's have a look at the main machine learning methods used to create Deepfakes are based on deep learning and involve creating generative neural network architecture, such as auto encoders or generative adversarial networks. That is GAN, Deepfakes have garnered widespread attention for their use in celebrity, bold videos, revenge, fake news and financial fraud.

Now, let's have a look at the real time implementation of Deepfake. 

In the example, you can see that it's a Deepfake technology we use in a scene from man of steel actress Amy Adams in the original left. And it's modified, to have the face of actor Nicholas Cage, So this is how the Deepfake actually, this is an interesting application. So you can use this application for some better development as well.

Now, let's have a look at the most popular Deepfake. That is USA president, Donald Trump.

You can see in the image that Donald Trump's face is actually put into some other one’s body, right? So this is how our Deepfake technology works. So this has Better Call Trump.

Why we are using Deepfakes?

The technology offers interesting possibilities, which include dubbing, improving and repairing video content, solving the Uncanny Valley effect in the video games, avoiding actors having to repeat fluffed lines or mistakes. The creation of apps that allow people to try clothes or different hairstyles and train doctors. So these are the most useful uses of Deepfake. We can make our life easy by using Deepfake technology.

Now let's have a look at some popular deepfakes. You can see the Bilawal Bhutto actually replaced by the Deefake. This is quite funny, right?

 

Disadvantages of using Deepfake Technology

Now, we will have a look at the nefarious uses of Defakes that are actually harming our society, like misusing celebrity images for explicit adult content, misusing political figures image for propagating false propagandas and financial crimes. These are all negative part of the Deepfake technology.

How are Deepfakes made?

You can see that Donald Trump most useful or most popular Deepfake of Donald Trump. So basically in the Deepfake, what we do, we use a generative adversarial network. So this network has two parts. One is Generative part another is the Discriminative part.So what actually it does? So in this case, we put some noise to our generator and generator does not have any idea about the actual image and in the discriminator we put the both images, the generator generated images and the actual image. So for that our discrimination can say this is a fake image or a genuine image, right ? With time we train our generator and discriminator both. and our generator and discriminator become this much efficient enough that our generator able to make an image like a real one. So if you don't see the person before, you can say that maybe this is an actual image, right? So GAN actually makes us fool. So in that case, our discriminator becomes that much efficient to say, no, whatever the real image like a replica of real image you try to make, but this is still fake. The actual images is the actual image you have provided.

So this is how our Deepfake technology work.