arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2102.10485cs.CVcs.AI
收藏

基于图像数据的条件生成对抗神经网络的可扩展平衡训练

Scalable Balanced Training of Conditional Generative Adversarial Neural Networks on Image Data

Oak Ridge National Laboratory(橡树岭国家实验室);Politecnico di Milano(米兰理工大学);Anthem, Inc.(Anthem公司)

Massimiliano Lupo Pasini, Vittorio Gabbi, Junqi Yin, Simona Perotto, Nouamane Laanait

摘要:

我们提出了一种分布式方法来训练深度卷积生成对抗性神经网络(DC-CGAN)模型。我们的方法通过根据数据标签划分训练数据来减少生成器和训练器之间的不平衡,并通过执行并行训练来增强可扩展性,其中多个生成器同时训练,其中每个生成器都专注于单个数据标签。在MNIST,CIFAR 10,CIFAR 100和ImageNet 1 k数据集上的初始分数和图像质量方面评估了性能,与训练DC-CGAN的最先进技术相比,显示出显着的改进。在OLCF超级计算机Summit上使用多达1,000个进程和2,000个NVIDIA V100 GPU,在所有四个数据集上实现弱缩放。

英文摘要:

We propose a distributed approach to train deep convolutional generative adversarial neural network (DC-CGANs) models. Our method reduces the imbalance between generator and discriminator by partitioning the training data according to data labels, and enhances scalability by performing a parallel training where multiple generators are concurrently trained, each one of them focusing on a single data label. Performance is assessed in terms of inception score and image quality on MNIST, CIFAR10, CIFAR100, and ImageNet1k datasets, showing a significant improvement in comparison to state-of-the-art techniques to training DC-CGANs. Weak scaling is attained on all the four datasets using up to 1,000 processes and 2,000 NVIDIA V100 GPUs on the OLCF supercomputer Summit.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑