A gesture classification SoC for rehabilitation with ADC-less mixed-signal feature extraction and training capable neural network classifier

Keyphrases

• Neural Network 100%
• System-on-chip 100%
• Classification System 100%
• Analog-to-digital Converter 100%
• Neural Network Classifier 100%
• Mixed-signal 100%
• Gesture Classification 100%
• Signal Feature Extraction 100%
• Feature Training 100%
• Gait Recognition 66%
• Low-dimensional Features 66%
• Area Cost 66%
• Fully Integrated 33%
• Classification Accuracy 33%
• Data Communication 33%
• Specific Training 33%
• Low Power 33%
• Multi-chip 33%
• Network Bandwidth 33%
• Rehabilitation Application 33%
• Feature Data 33%
• Sensor Fusion 33%
• User-specific 33%
• Power Processes 33%
• Average Power 33%
• Time Domain Features 33%
• Analog Front-end 33%
• Network Layer 33%
• Multiple chips 33%
• Computational Latency 33%
• Fully Convolutional Neural Network 33%
• Power Consumption Cost 33%
• Channel Test 33%
• Communication Bottleneck 33%
• On-chip Learning 33%

Computer Science

• Analog-to-Digital Converter 100%
• Neural Network 100%
• Feature Extraction 100%
• System-on-Chip 100%
• Dimensional Feature 50%
• Power Consumption 25%
• Layer Neural Network 25%
• Communication Bandwidth 25%
• Data Communication 25%
• Classification Accuracy 25%
• Domain Feature 25%

Engineering

• Feature Extraction 100%
• System-on-Chip 100%
• Analog-to-Digital Converter 100%
• Time Domain 33%
• Electric Power Utilization 33%
• Classification Accuracy 33%
• Average Power 33%
• Front End 33%
• Domain Feature 33%
• Layer Network 33%
• Sensor Fusion 33%
• Analog Front 33%