Convolutional neural network pruning optimization method and device, electronic equipment and storage medium

Abstract

The invention provides a convolutional neural network pruning optimization method and apparatus, an electronic device and a storage medium. The method comprises the steps of obtaining a trained convolutional neural network model; for each category of the convolutional neural network model, determining a corresponding semantic information graph, and determining a filter importance factor of each filter in each category based on the semantic information graph; according to the importance factor of the filter and the pruning target, carrying out importance degree sorting on the filter; on the basis of the sorting result and a pruning target, the filters with the small importance degree are cut off step by step until the pruning target is achieved, and a convolutional neural network model after pruning optimization is obtained; and carrying out retraining on the convolutional neural network model after pruning optimization. According to the method, targeted convolutional neural network pruning compression with a better effect can be realized.

Description

technical field [0001] Embodiments of the present invention relate to the field of deep learning technology, and in particular to a convolutional neural network pruning optimization method, device, electronic equipment, and storage medium. Background technique [0002] Convolutional neural networks are very important in the field of artificial intelligence, and common applications include computer vision, speech recognition, and natural language processing. Convolutional neural networks can improve network performance by deepening the hierarchical structure of the network. By increasing the size of the neural network model, the learning task effect can be improved, but it also brings difficulties in the deployment of the convolutional neural network. The main problem is that there is huge redundancy in the internal parameters of the neural network model, resulting in waste of resources. Existing studies have shown that only a small subset of parameters can be given, and th...

AI Technical Summary

Problems solved by technology

[0004] Based on the problem in the prior art that the pruning optimization effect of the convolutional neural network is not good for general scenarios of the overall network, the embodiments of the present invention provide a targeted convolutional neural network pruning optimization method, device, and electronic equipment and storage media, which can combine categories to realize customized optimization of convolutional neural networks

Images