Bibliographic Details
| Title: |
An Introduction to Generative Adversarial Nets in R: The RGAN Package. |
| Authors: |
Neunhoeffer, Marcel1,2 marcel@marcel-neunhoeffer.com |
| Source: |
R Journal. Mar2026, Vol. 18 Issue 1, p5-22. 18p. |
| Subject Terms: |
*Generative adversarial networks, *Software libraries (Computer programming), *Machine learning, *Synthetic data, *Programming languages, *Artificial neural networks |
| Abstract: |
This article introduces Generative Adversarial Nets (GAN), a powerful neural net architecture, and presents the RGAN package. RGAN makes it easy to implement GANs in R: It facilitates experimentation with different design choices for GANs, such as value functions, other network architectures, the choice of different noise distributions, the choice of optimization algorithms and update schemes, dropout during data generation and the post-processing of generated GAN samples. Furthermore, advanced users can provide customized functions for each design choice. RGAN is a lightweight package and runs on top of the torch library in R. [ABSTRACT FROM AUTHOR] |
| Database: |
Academic Search Index |