Comprehensive evaluation of openCL-based CNN implementations for FPGAs

Mathematics

• Comprehensive Evaluation 100%
• Field Programmable Gate Array 88%
• Neural Networks 52%
• Architecture 46%
• Inconsistency 32%
• Co-design 27%
• Resources 26%
• GPGPU 26%
• Many-core 25%
• Memory Hierarchy 25%
• Visual Cortex 21%
• Network Architecture 21%
• Hardware Implementation 21%
• Shared Memory 20%
• Execution Time 18%
• Parallel Computation 17%
• Design 17%
• Hierarchical Structure 17%
• Hardware 17%
• High Power 17%
• Artificial Intelligence 16%
• Performance 16%
• Parallelism 16%
• Real-time 15%
• Classifier 15%
• Industry 14%
• Community 14%
• Communication 12%
• Learning 12%
• Synthesis 12%
• Convolution 11%
• Unit 9%
• Demonstrate 8%
• Framework 8%

Engineering & Materials Science

• Convolutional neural networks 51%
• Field programmable gate arrays (FPGA) 45%
• Data storage equipment 34%
• Deep neural networks 18%
• Cache memory 12%
• Computer hardware 10%
• Convolution 9%
• Network architecture 8%
• Graphics processing unit 8%
• Deep learning 8%
• Artificial intelligence 8%
• Program processors 7%
• Classifiers 6%
• Communication 4%
• Industry 4%