Evaluating POWER Architecture for Distributed Training of Generative Adversarial Networks

INIS

• power 100%
• scaling 100%
• architecture 100%
• performance 60%
• data 60%
• datasets 40%
• resources 40%
• efficiency 20%
• evaluation 20%
• neural networks 20%
• learning 20%
• availability 20%
• constraints 20%

Computer Science

• Generative Adversarial Networks 100%
• Computing Resource 66%
• Data Parallelism 33%
• Training Process 33%
• Large Data Set 33%
• Data Scientist 33%
• Deep Neural Network 33%
• Training Phase 33%
• Driven Approach 33%
• High-Performance Computing 33%

Chemical Engineering

• Deep Learning 100%
• Deep Neural Network 100%