IR-QNN Framework: An IR Drop-Aware Offline Training Of Quantized Crossbar Arrays

Engineering

• SPICE 100%
• Offline Training 100%
• Design Consideration 50%
• Resistive 50%
• Case Scenario 50%
• Deep Neural Network 50%
• Driver Circuit 50%
• Defects 50%
• Interconnects 50%
• Realization 50%
• Perceptron 50%

Computer Science

• Deep Neural Network 100%
• Mapping Scheme 33%
• Vector Multiplication 33%
• Associative Processor 33%
• Estimation Method 33%
• Design Consideration 33%
• Training Process 33%
• Multilayer Perceptron 33%

Keyphrases

• IR Drop 100%
• Interconnect Wire 25%
• Stuck-at 25%
• VGGNet 25%
• Quantized Weight 25%
• Deep Neural Network Acceleration 25%