Data Structure
Networking
RDBMS
Operating System
Java
MS Excel
iOS
HTML
CSS
Android
Python
C Programming
C++
C#
MongoDB
MySQL
Javascript
PHP
Physics
Chemistry
Biology
Mathematics
English
Economics
Psychology
Social Studies
Fashion Studies
Legal Studies
- Selected Reading
- UPSC IAS Exams Notes
- Developer's Best Practices
- Questions and Answers
- Effective Resume Writing
- HR Interview Questions
- Computer Glossary
- Who is Who
Use Cases of Generative Adversarial Networks
Introduction
In recent years Generative Adversarial Networks have gained a lot of traction and used in numerous use cases. GANs are deep neural network architectures used to generate realistic synthetic data from existing data sets. GANs have two sets of neural networks − the first one generates new images and the second one discriminates between real and generated images. The discriminants give feedback to the generator about the image whether it is real or fake so that the generator can produce better images that look exactly like the real ones. This process is continued till the discriminator fails to differentiate between real and fake images.
In this article let us study some of the use cases of Generative Adversarial Networks.
Use Cases of Generative Adversarial Networks
Generating Realistic Image Datasets − GANs are used to generate real−looking image datasets for different domains like space technology, medical images, etc. This helps Machine Learning Engineers/researchers to use this data to improve the performance of existing models by using data augmentation with generated images.
Generating human photos and animated characters − GANs can generate exact image photos of real people as well as create images of people that do not exist. These images can be used as avatars for various platforms like gaming, and social media. They can also generate high−resolution cartoon characters from movie characters. These characters can be used in highly engaging games to provide a real experience to the user.
Aging of Face − GANs can help create aged images of people in different age groups. This can help users to see the effect of aging on them.
Improving Resolution of Images (Super Resolutions) − GANs can enhance the resolution of pixelated photographs thus increasing the quality of images.
Generating images from text − GANs are used to generate images from captions and text descriptions alone. They can also create photos from scenic descriptions and any other feature or attribute description.
Synthesis of Videos − GANs are used to generate future sequences of scenes of human activity, animated scenes, or any other video scene.
Generating music − GANs can be used to create new and realistic music instrument tunes and genres.
3D model creation − GANs have been used to create realistic 3D models suitable for various 3D scenes. The creation can take place from existing data or even 2D images. These 3D images are used in Virtual Reality and Computer Aided Design(CAD).
Human Pose Generation − GANs can produce human poses that are impossible to perform by humans.
Detection of Anomalies − GANs can be used to identify anomalies in data that include fraud detection, medical anomalies, etc
Privacy and Security − GANs can prevent cyber threats and adversarial attacks. Hackers use fake data to fool the system. Thus GANs can be made to create fake data and learn the patterns and prevent such threats in the future.
Conclusion
There are a plethora of use cases for GANs. They have shown us how beautifully we can use deep learning and AI to create content that is not only realistic but sometimes difficult for even humans to distinguish between and fake. Due to this nature, GANs have proved used in many situations where similar issues are found.
58 Views
