On the impact of deep neural network calibration on adaptive edge offloading for image classification

Keyphrases

• Accuracy Requirements 20%
• Adaptive Edge 100%
• Adaptive Offloading 20%
• Calibration Study 20%
• Classification Task 20%
• Communication Delay 20%
• Confidence Estimation 20%
• Confidence Level 40%
• Deep Neural Network 100%
• Deep Neural Network Inference 20%
• Different Datasets 20%
• Early Exit 100%
• Edge Computing 20%
• Edge Devices 20%
• Edge Task Offloading 100%
• Energy Conservation 20%
• Image Classification 100%
• Inference Time 20%
• Latency 20%
• Latency Constraint 20%
• Network Calibration 100%
• Network Layer 20%
• Offloading Decision 40%
• Prediction Confidence 20%
• Processing Constraints 20%
• Side Branch 60%

Engineering

• Application Level 25%
• Classification Task 25%
• Confidence Level 50%
• Deep Neural Network 100%
• Edge Computing 25%
• Energy Constraint 25%
• Image Classification 100%
• Intermediate Layer 25%
• Metrics 25%
• Network Model 25%
• Side Branch 75%

Computer Science

• Application Level 11%
• Classification Task 11%
• Communication Delay 11%
• Confidence Level 22%
• Deep Neural Network 100%
• Edge Computing 11%
• Image Classification 100%
• Network Inference 11%
• Network Layer 11%
• Neural Network Model 11%