Compression of neural networks for specialized tasks via value locality

Computer Science

• Neural Network 100%
• Convolutional Neural Network 100%
• Experimental Result 40%
• Training Process 40%
• Compression Algorithm 40%
• Activation Function 40%
• Deep Neural Network 20%
• Image Classification 20%
• Neural Network Model 20%
• Systems Performance 20%
• Inference Process 20%
• Energy Consumption 20%
• Preprocessing Stage 20%
• Computational Resource 20%
• Computer Vision 20%
• Output Feature Map 20%
• Compression Stage 20%

Keyphrases

• Neural Network 100%
• Value Locality 100%
• Locality-based 69%
• Convolutional Neural Network 30%
• Activation Function 15%
• Training Process 15%
• Compression Algorithms 15%
• Compress 15%
• Process Value 15%
• Image Classification 7%
• Deep Neural Network 7%
• Computational Resources 7%
• Computer Vision 7%
• Cost Consumption 7%
• Computationally Intensive 7%
• Preprocessing Techniques 7%
• Computer Image 7%
• Inference Process 7%
• Energy Consumption 7%
• Computational Requirements 7%
• System Energy 7%
• Feature Map 7%
• Performance System 7%
• Edge Devices 7%
• Network Value 7%
• Output Features 7%
• Saving Rate 7%
• Intensive Training 7%
• Convolutional Neural Network Model 7%
• System Cost 7%
• Neural Network Compression 7%
• Network Task 7%
• Compressor Stage 7%