EdgeL³: Compressing L³-Net for mote scale urban noise monitoring

Keyphrases

• Mote 100%
• Urban Noise Monitoring 100%
• Edge Devices 66%
• Knowledge Distillation 66%
• Audio-visual Correspondence 66%
• Classification Task 33%
• Order of Magnitude 33%
• Internet of Things 33%
• Popular 33%
• Resource-constrained 33%
• Training Data 33%
• Long-lived 33%
• Transfer Learning Model 33%
• Microcontroller 33%
• Pruning 33%
• Floating-point Representation 33%
• Ablation 33%
• Compression Strategy 33%
• Embedded Devices 33%
• Distillation Technique 33%
• Limited Memory 33%
• Constrained Edge 33%
• Audio Classification 33%
• Urban Noise 33%
• Self-powered Operation 33%
• Small Edges 33%
• Noise Sensing 33%
• Sound Classification 33%
• Deep Audio Embeddings 33%
• Half-precision Floating Point 33%

Computer Science

• Knowledge Distillation 100%
• Classification Task 50%
• Internet of Things 50%
• Reference Model 50%
• Learning Technique 50%
• Training Data 50%
• Transfer Learning 50%
• Convolutional Neural Network 50%
• Floating Point 50%
• Point Representation 50%
• Microcontroller 50%
• Embedded Device 50%
• Audio Classification 50%