Improving Reliability of ReRAM-Based DNN Implementation through Novel Weight Distribution

Mathematics

• Random Access 100%
• Weight Distribution 86%
• Hardware 52%
• Deviation 38%
• Neural Networks 27%
• Grain Size 22%
• Binary 21%
• Unit 19%
• Degradation 17%
• Performance 17%
• Fault 16%
• Architecture 15%
• Baseline 15%
• Mean Value 14%
• Cell 10%
• Simulation 9%

Engineering & Materials Science

• Data storage equipment 43%
• Computer hardware 31%
• Memory architecture 22%
• Static random access storage 19%
• Deep neural networks 17%
• Degradation 12%
• Neural networks 10%