Model compression methods for convolutional neural networks

Keyphrases

• Compression Method 100%
• Convolutional Neural Network 100%
• Model Compression 100%
• Latency 50%
• Dropout 50%
• Computer Vision 50%
• Fully Connected Layer 50%
• Convolutional Layer 50%
• Feature Map 50%
• Pruning 25%
• Effective Solutions 25%
• Mobile Phone 25%
• Embedded Platform 25%
• Overfitting 25%
• Desktop 25%
• Memory Bandwidth 25%
• Model Size 25%
• Extracting Features 25%
• Vehicle Classification 25%
• Points Reduction 25%
• Deep Learning 25%
• Processing Resources 25%
• CPU-GPU 25%
• Mobile IoT 25%
• Convolutional Networks 25%
• Embedded Target 25%
• Vehicle Classifier 25%
• Inference Time 25%
• IoT Edge Computing 25%
• Sparse Formats 25%
• Pruned Models 25%
• Memory Cache 25%
• Sparse Matrix Format 25%

Computer Science

• Model Compression 100%
• Convolutional Neural Network 100%
• Convolutional Layer 50%
• Graphics Processing Unit 50%
• Computer Vision 50%
• Fully Connected Layer 50%
• Feature Map 50%
• Effective Solution 25%
• Embedded Platform 25%
• Memory Bandwidth 25%
• Regularization 25%
• Convolutional Network 25%
• Processing Resource 25%
• Component Vector 25%
• Cache Memory 25%
• Internet-Of-Things 25%
• Deep Learning Method 25%