09.05.2023 Views

pdfcoffee

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Generative Adversarial Networks

One of the coolest uses of GANs is doing arithmetic on faces in the generator's

vector Z. In other words, if we stay in the space of synthetic forged images, it is

possible to see things like this: [smiling woman] - [neutral woman] + [neutral man]

= [smiling man], or like this: [man with glasses] - [man without glasses] + [woman

without glasses] = [woman with glasses]. This was shown in the paper Unsupervised

Representation Learning with Deep Convolutional Generative Adversarial Networks by

Alec Radford and his colleagues in 2015. All images in this work are generated by

a version of GAN. They are NOT REAL. The full paper is available here: http://

arxiv.org/abs/1511.06434. Following are some examples from the paper. The

authors also share their code in this GitHub repo: https://github.com/Newmu/

dcgan_code:

[ 216 ]

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!