Conditional Generative Adversarial Network
Conditional Generative Adversarial Networks (cGANs) are a type of generative model designed to create new data samples conditioned on specific input information, aiming to generate realistic and diverse outputs matching given constraints. Current research focuses on applying cGANs to diverse fields, leveraging architectures like Pix2Pix and CycleGAN, and exploring variations such as those incorporating diffusion models or transformer networks for improved performance and stability. This approach holds significant promise for various applications, including medical imaging (e.g., synthesizing missing data or enhancing image quality), manufacturing (generating manufacturable designs), and other domains requiring data augmentation or realistic data synthesis where obtaining real data is difficult or expensive. The ability to generate high-quality synthetic data addresses limitations in data availability and privacy concerns, ultimately improving the performance and generalizability of downstream machine learning models.
Papers
DexGANGrasp: Dexterous Generative Adversarial Grasping Synthesis for Task-Oriented Manipulation
Qian Feng, David S. Martinez Lema, Mohammadhossein Malmir, Hang Li, Jianxiang Feng, Zhaopeng Chen, Alois Knoll
McGAN: Generating Manufacturable Designs by Embedding Manufacturing Rules into Conditional Generative Adversarial Network
Zhichao Wang, Xiaoliang Yan, Shreyes Melkote, David Rosen
Applying Conditional Generative Adversarial Networks for Imaging Diagnosis
Haowei Yang, Yuxiang Hu, Shuyao He, Ting Xu, Jiajie Yuan, Xingxin Gu
Enhancing the Utility of Privacy-Preserving Cancer Classification using Synthetic Data
Richard Osuala, Daniel M. Lang, Anneliese Riess, Georgios Kaissis, Zuzanna Szafranowska, Grzegorz Skorupko, Oliver Diaz, Julia A. Schnabel, Karim Lekadir