FracBNN: Accurate and fpga-efficient binary neural networks with fractional activations

Keyphrases

• Fpga 100%
• Binary Neural Network 100%
• Neural Network Model 22%
• ImageNet 22%
• Binarization 22%
• Model Size 22%
• Input Layer 22%
• Embedded FPGA 22%
• Thermometer 11%
• Network Architecture 11%
• 1-bit 11%
• Memory Requirements 11%
• Resource-constrained 11%
• Optimized Network 11%
• Xilinx 11%
• FPGA Device 11%
• FPGA Acceleration 11%
• Realistic Dataset 11%
• Neural Network Design 11%
• Bitwise 11%
• Accuracy Loss 11%
• Convolutional Neural Network Model 11%
• Convolutional Layer 11%
• MobileNetV2 11%
• Dual-precision 11%
• Bit Weight 11%
• Activation Schemes 11%
• MAC Operations 11%
• Real-time Image Classifications 11%
• Binary Convolution 11%
• Compact Convolutional Neural Networks 11%

Computer Science

• Neural Network 100%
• Field Programmable Gate Arrays 71%
• Neural Network Model 42%
• Network Architecture 14%
• Memory Requirement 14%
• Network Design 14%
• Image Classification 14%
• binarization 14%
• Convolutional Layer 14%
• Convolutional Neural Network 14%
• MobileNetV2 14%

Engineering

• Field Programmable Gate Arrays 100%
• Network Model 60%
• Input Layer 40%
• Memory Requirement 20%
• Image Classification 20%
• Convolutional Layer 20%
• Convolutional Neural Network 20%