Low Cost Convolutional Neural Network Accelerator Based on Bi-Directional Filtering and Bit-Width Reduction

Keyphrases

• Accuracy Degradation 33%
• Activation Sparsity 33%
• AlexNet 33%
• Area Saving 33%
• Area-efficient 66%
• Bidirectional Filter 100%
• Bit-width 33%
• Bit-width Optimization 100%
• Cellular Layout 33%
• Computational Cost 33%
• Convolutional 33%
• Convolutional Neural Network 66%
• Convolutional Neural Network Accelerator 100%
• Data Movement 33%
• Data Redistribution 33%
• Design Techniques 66%
• Differential Input 66%
• Energy Efficiency 33%
• Energy Saving 33%
• Energy-efficient Hardware 33%
• Fast Data 33%
• Filtering Operation 33%
• Filtering Window 33%
• Hardware Accelerator 33%
• Hardware Cost 33%
• Ignorable 33%
• Input Method 66%
• Low Area 33%
• Low Energy 33%
• Memory Access 33%
• Multiply-accumulate 33%
• Near-zero 33%
• Network Accuracy 33%
• Numerical Results 33%
• Output-only 33%
• Processing Time 33%
• Reduction Strategies 33%
• SRAM Cell 33%
• Zero-skipping 33%

Engineering

• Computational Cost 16%
• Convolutional Neural Network 100%
• Design Technique 33%
• Differential Input 33%
• Directional 100%
• Energy Conservation 33%
• Energy Efficiency 16%
• Energy Engineering 33%
• Filtration 100%
• Hardware Accelerator 16%
• Hardware Cost 16%
• Memory Access 16%
• Processing Time 16%
• Sparsity 16%