Efficient Software-Implemented HW Fault Tolerance for TinyML Inference in Safety-critical Applications

Keyphrases

• Fault Tolerance 100%
• Safety-critical Applications 100%
• TinyML 100%
• Efficient Software 100%
• Runtime Overhead 60%
• Microcontroller Unit 40%
• Soft Error 40%
• Depthwise Separable Convolution 40%
• Adaptation 20%
• Drone 20%
• Low Power 20%
• Design Challenges 20%
• Resilience 20%
• Soft Error Resilience 20%
• Fully Connected Layer 20%
• Code Size 20%
• Fully Convolutional Neural Network 20%
• Energy Use 20%
• Efficient Execution 20%
• Fault-tolerance Techniques 20%
• Algorithm-based Fault Tolerance 20%
• Epilogue 20%
• Selective Hardening 20%
• Unmanned Autonomous Vehicles 20%
• Misprediction 20%
• Layerwise Theory 20%
• MLPerf 20%
• Latency Coding 20%
• Neural Network Inference 20%

Computer Science

• fault-tolerance 100%
• Convolutional Neural Network 100%
• Soft Error 100%
• Critical Application 100%
• Neural Network 66%
• Micro Controller Unit 66%
• Error Resilience 33%
• Network Inference 33%
• Autonomous Vehicles 33%
• Design Challenge 33%
• Efficient Execution 33%
• Tolerance Technique 33%
• Experimental Result 33%
• Fully Connected Layer 33%