μl2Q: An Ultra-Low Loss Quantization Method for DNN Compression

Keyphrases

• Quantization Method 100%
• Deep Neural Network 100%
• Ultra-low Loss 100%
• Deep Neural Network Compression 100%
• Effective Method 40%
• Network Accuracy 40%
• Bit-width 40%
• Data-free Quantization 40%
• Popular 20%
• Compress 20%
• Compression Ratio 20%
• Accuracy Improvement 20%
• Computing Capabilities 20%
• Energy Consumption 20%
• Memory Footprint 20%
• Standard Normal Distribution 20%
• Intermediate Data 20%
• Quantization Scheme 20%
• Deep Neural Network Training 20%
• Accuracy Loss 20%
• Machine Learning Framework 20%
• Quantitative Data Analysis 20%
• Neural Network Quantization 20%
• Deep Neural Network Design 20%
• Caffe 20%
• End-to-end Deep Learning 20%
• ResNet18 20%
• LeNet-5 20%

Computer Science

• Quantization (Signal Processing) 100%
• Deep Neural Network 100%
• Effective Method 22%
• Machine Learning 11%
• Compression Ratio 11%
• Neural Network Model 11%
• Quantitative Data 11%
• Normal Distribution 11%
• Memory Footprint 11%
• Energy Consumption 11%
• Network Design 11%
• Learning Framework 11%
• Intermediate Data 11%

Engineering

• Quantization (Signal Processing) 100%
• Deep Neural Network 100%
• Energy Engineering 11%
• Network Model 11%
• Space Data 11%
• Compression Ratio 11%
• Accuracy Improvement 11%
• Memory Footprint 11%
• Standard Normal Distribution 11%