Failure-Resilient ML Inference at the Edge through Graceful Service Degradation

Keyphrases

• Machine Learning Models 100%
• Service Degradation 100%
• Inference at the Edge 100%
• Resilient Machine Learning 100%
• Machine Learning Inference 100%
• Failover 66%
• Computational Cost 33%
• Potential Gains 33%
• Sampling Rate 33%
• Constrained Edge 33%
• Network Device 33%
• System Reliability 33%
• Classification Decision 33%
• Resource-constrained 33%
• Crash Failure 33%
• Potential Trade 33%
• Accuracy Rate 33%
• Edge Habitat 33%
• Deep Learning 33%
• Latency-sensitive 33%
• Infrastructure Cost 33%
• Byzantine Failures 33%
• Edge Resources 33%
• Classification Detection 33%
• Cost Overhead 33%
• IoT Devices 33%
• IoT Applications 33%
• Latency Rate 33%
• Machine Learning Technology 33%
• Edge Devices 33%
• Recent Innovations 33%
• Various Approaches 33%
• Traditional Solution 33%

Computer Science

• Machine Learning 100%
• Learning System 100%
• Machine Learning Model 100%
• Decision-Making 33%
• Sampling Rate 33%
• Future Direction 33%
• Potential Benefit 33%
• Latency Rate 33%
• IoT Application 33%
• Internet-Of-Things 33%
• Deep Learning Method 33%
• Machine Learning Technology 33%