You are hired by a video game company as a data scientist. The company is facing challenges and needs you to save their business.

The game is famous for its unique characters for every player. With the growth of the player amount, it comes to be a nearly impossible mission for the artist to hand plot the characters for millions of players. But your boss plans to keep the unique character-creating function in the game to keep the customers.

How can we mass-produce anime characters using a machine-learning method?
You will create anime characters like the ones below, using the DCGANs model in this guided project.

face_cartton.png 465 KB

As a data scientist, you know that Generative adversarial networks (GAN) can help with the task.
GANs is a class of machine learning frameworks, which could generate new and realistic photograph that is authentic to human observers. Moreover, applying the Convolutional networks (CNNs) to GANs models could facilitate the photo generating model. The combined method is called Deep Convolutional Generative Adversarial Networks (DCGANs). 

You are going to train a DCGANs model, using the existing character, for the further massive unique anime characters production for the video game.

In the guided project, you will first learn the basic about GANs, using toy data to understand what generator and discriminator is.
In the second half of the guided project, you will train Deep Convolutional Generative Adversarial Networks (DCGANs) models to create anime characters.
After the guided project, you will be able to:

This course is good for intermediate learners in Machine Learning and Data Science. 
Knowing basic Python usage in data science is suggested before you start this guided project. 
We recommend using the IBM Skills Network Labs environment for this guided project. Everything you need to complete this project will be provided to you via the Skills Network Labs. The platform works best with current versions of Chrome, Edge, Firefox, Internet Explorer or Safari.