Retrain or not retrain? -- efficient pruning methods of deep CNN networks
This addresses the challenge of model compression for image processing tasks, but it appears incremental as it compares existing pruning methods.
The paper tackles the problem of efficiently pruning deep convolutional neural networks (CNNs) to reduce complexity and memory footprint, comparing retraining and no-retraining approaches to find near-optimal solutions with specified accuracy drops.
Convolutional neural networks (CNN) play a major role in image processing tasks like image classification, object detection, semantic segmentation. Very often CNN networks have from several to hundred stacked layers with several megabytes of weights. One of the possible methods to reduce complexity and memory footprint is pruning. Pruning is a process of removing weights which connect neurons from two adjacent layers in the network. The process of finding near optimal solution with specified drop in accuracy can be more sophisticated when DL model has higher number of convolutional layers. In the paper few approaches based on retraining and no retraining are described and compared together.