DEGAN: Decentralized generative adversarial networks

Mohammad Hashem Faezi | Shahriar Bijani | Ardeshir Dolatimalekabad

---

URL :   http://research.shahed.ac.ir/WSR/WebPages/Report/PaperView.aspx?PaperID=137811
Date :  2021/01/20
Publish in :    Neurocomputing
DOI :  https://doi.org/https://doi.org/10.1016/j.neucom.2020.07.089
Link :  http://dx.doi.org/https://doi.org/10.1016/j.neucom.2020.07.089
Keywords :Deep learning, Generative adversarial networks, Distributed machine learning, Decentralized architecture

Abstract :

---

We propose a distributed and decentralized Generative Adversarial Networks (GANs) framework without the exchange of the training data. Each node contains local dataset, a discriminator and a generator, from which only the generator gradients are shared with other nodes. In this paper, we introduce a novel, distributed technique in which workers communicate directly with each other, having no central nodes. Our experimental results on the benchmark datasets demonstrate almost the same performance and accuracy compared with existing centralized GAN frameworks. The proposed framework addresses the lack of decentralized learning for GANs.