Where to Prune: Using LSTM to Guide Data-Dependent Soft Pruning

Computer Science

• Accuracy 33%
• Baseline Model 33%
• Channels 66%
• Complexity Reduction 33%
• Compression Rate 33%
• Computational Complexity 33%
• Computational Cost 33%
• Convolutional Neural Network 100%
• Embedded Device 33%
• Experimental Result 33%
• Floating-Point Operation 33%
• Global Network 33%
• Long Short-Term Memory Network 66%
• Model Accuracy 33%
• Networks 33%
• Neural Network Model 33%
• Storage Overhead 33%