Refresh Power Reduction of DRAMs in DNN Systems Using Hybrid Voting and ECC Method

Keyphrases

• Neural Network System 100%
• Dynamic Random Access Memory 100%
• Deep Neural Network 100%
• Refresh Energy 100%
• Error Correction Codes 100%
• Power Reduction 100%
• Refresh Interval 66%
• Weighted Data 66%
• Power Saving 33%
• U-Net 33%
• Code-based 33%
• Data Retention Fault 33%
• AlexNet 33%
• Bit Overhead 33%
• Data Buffering 33%
• Accuracy Loss 33%
• ImageNet 33%
• Check Bits 33%
• Period Extension 33%
• Convolutional Neural Network (CNN) Model 33%
• VGG19 33%

Engineering

• Neural Network System 100%
• Dynamic Random Access Memory 100%
• Correction Code 100%
• Error Correction 100%
• Deep Neural Network 100%
• Data Retention 25%
• Check Bit 25%
• Convolutional Neural Network 25%