Learning Task-Oriented Communication for Edge Inference: An Information Bottleneck Approach

Keyphrases

• Learning Task 100%
• Task Engagement 100%
• Information Bottleneck 100%
• Encoded Features 100%
• Edge Inference 100%
• Communication Systems 66%
• Feature Vector 66%
• Dynamic Channel Conditions 66%
• Variational Information Bottleneck 66%
• Rate-distortion Tradeoff 66%
• Variational 33%
• Feature Extraction 33%
• Task-oriented 33%
• Adaptive Adjustment 33%
• Inference Performance 33%
• Sparsity 33%
• Communication Overhead 33%
• Local Data 33%
• Limited Bandwidth 33%
• Communication Scheme 33%
• Learning-based 33%
• High-dimensional Data 33%
• Channel Condition 33%
• Source Code 33%
• Channel Coding 33%
• Edge Server 33%
• Edge Devices 33%
• Dynamic Neural Network 33%
• Inference Tasks 33%
• Feature Transmission 33%
• Data Reconstruction 33%
• Compact Representation 33%
• Transmission Latency 33%
• Variational Approximation 33%
• Low-latency Inference 33%
• Lower End 33%
• Bottleneck Optimization 33%
• Feature Encoding Scheme 33%

Computer Science

• Channel Condition 100%
• rate distortion 66%
• Feature Vector 66%
• High Dimensional Data 33%
• Sparsity 33%
• Edge Server 33%
• Communication Overhead 33%
• Communication Scheme 33%
• Channel Coding 33%
• Baseline Method 33%
• Approximation (Algorithm) 33%
• Inference Task 33%
• Neural Network 33%
• Encoding Scheme 33%
• Extracted Feature 33%
• Variable Length 33%
• Compact Representation 33%
• Feature Extraction 33%