AppCiP: Energy-Efficient Approximate Convolution-in-Pixel Scheme for Neural Network Acceleration

Keyphrases

• Energy Efficient 100%
• Power Consumption 100%
• Neural Network Acceleration 100%
• Always-on 50%
• High Efficiency 50%
• Unique Capabilities 50%
• Order of Magnitude 50%
• Low Speed 50%
• Neural Network 50%
• Highly Parallel 50%
• Artificial Intelligence 50%
• Art Design 50%
• Feature Extraction 50%
• Sensing Device 50%
• Limited Resources 50%
• Energy-intensive 50%
• Memory Bottleneck 50%
• Reconfigurable 50%
• Face Detection 50%
• Different Datasets 50%
• Performance Accuracy 50%
• Cloud Processors 50%
• Low-precision 50%
• Self-powered 50%
• Area Overhead 50%
• Floating-point 50%
• Long Latency 50%
• Analog-to-digital Converter 50%
• Limited Sensing 50%
• Digital-to-analog 50%
• CIFAR-10 50%
• Co-simulation 50%
• Analog Buffer 50%
• Visual Perception System 50%
• Quinary 50%
• Integration Design 50%
• Pixel Architecture 50%
• Grayscale Conversion 50%

Computer Science

• Energy Efficient 100%
• Power Consumption 100%
• Neural Network 100%
• Artificial Intelligence 50%
• Feature Extraction 50%
• Convolutional Neural Network 50%
• Floating Point 50%
• Face Detection 50%
• Analog-to-Digital Converter 50%
• Capturing Data 50%