Image incremental learning method based on a migration model

Abstract

An image increment learning method based on a variational auto-encoder comprises the following steps: 1) constructing a main network taking a ResNet50 network layer structure as a prototype, and introducing an MMD distance metric; 2) setting an optimizer and a learning rate, and adopting a knowledge distillation and weight punishment strategy; 3) training the ResNet 50 model by adopting a restricted sample increment method in the training mode; and 4) reloading the optimal model trained recently, and repeatedly using the restrictive sample increment method for training until all increment dataare trained, so that the generation of the disturbance resistance is not limited by the influence of a plurality of environmental factors in practice, and the practical value is higher.

Description

technical field [0001] The present invention relates to an incremental learning method and a knowledge distillation technology, which uses the idea of ​​Transfer Learning for reference, uses the technique of Weight Punish, and combines the restrictive sample incremental learning method to control the training in each learning process The number of samples, under the premise of maintaining the classification and recognition accuracy of the old category, at the same time improve the classification accuracy of the new category data, so as to achieve incremental learning on the original model. Background technique [0002] In recent years, Deep Convolutional Neural Networks (DCNNs) have become the main structure for large-scale image classification. In 2012, AlexNet won the ImageNet Large Scale Visual Recognition Challenge (ISLVRC) by implementing Deep-CNN and pushing DCNNs into the public eye. Since then, they have dominated ISLVRC and performed well on popular image datasets ...

AI Technical Summary

Problems solved by technology

[0006] At present, the research on incremental learning technology at home and abroad is still in its infancy.

Images