Accurate deep neural network inference using computational phase-change memory

Keyphrases

• Phase Change Random Access Memory (PCRAM) 100%
• Computational Phase Change 100%
• Deep Neural Network Inference 100%
• Resistive Memory Devices 50%
• CIFAR-10 50%
• U-Net 50%
• In-memory Computing 25%
• Postsynaptic Density Protein 95 (PSD-95) 25%
• Convolutional Neural Network 25%
• Compensation Method 25%
• Classification Accuracy 25%
• Significant Loss 25%
• Non-von Neumann 25%
• Device Variability 25%
• ImageNet 25%
• Deep Learning Inference 25%
• Accuracy Loss 25%
• Device Noise 25%
• Accuracy Retention 25%
• Energy-efficient Deep Learning 25%
• Two-phase Deformation 25%
• Batch Normalization 25%
• Differential Configuration 25%

Computer Science

• Deep Neural Network 100%
• Network Inference 100%
• Resistive Memory 100%
• Residual Neural Network 100%
• Energy Efficient 50%
• Deep Learning 50%
• Classification Accuracy 50%
• Convolutional Neural Network 50%
• Synaptic Weight 50%
• Batch Normalization 50%