Neural network architecture search method based on convolution kernel prediction

Abstract

The invention provides a neural network architecture search method based on convolution kernel prediction. The method comprises the steps of: constructing a super network for neural network architecture search, wherein the super network comprises a teacher network and a student network, the teacher network is a pre-trained network, and the student network is composed of a plurality of basic units;training the student network by taking a training set and the coding information of a neural network architecture as input, and predicting to obtain an optimal convolution kernel when a loss functionconverges to a minimum value, wherein the loss function is determined according to the loss function of the teacher network and the loss function of the student network; and updating the coding information of the neural network architecture according to the loss function of the student network to obtain an optimal neural network architecture. According to the neural network architecture search method based on convolution kernel prediction, the teacher network is introduced as guidance, and a convolution kernel prediction module in the student network can accurately predict the optimal convolution kernel, so that the search efficiency is greatly improved, and the global optimum of the search result can be ensured.

Description

technical field [0001] The invention relates to the technical field of neural network architecture search, in particular to a neural network architecture search method based on convolution kernel prediction. Background technique [0002] Designing an efficient neural network architecture is an important research content in deep learning. An excellent neural network architecture can effectively improve the efficiency and accuracy of multiple tasks including image classification, detection, and segmentation, but manual design of network architecture is often serious. It relies on human experience, and at the same time creates a huge burden on the relevant researchers. Based on this, Neural Architecture Search (Neural Architecture Search, referred to as NAS) is proposed, and an optimal neural network architecture is obtained by automatically searching the neural network architecture in a predetermined search space. In many fields, such as image classification, image segmentati...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to provide a neural network architecture search method based on convolution kernel prediction to solve the technical problems of low search efficiency and unstable search results in neural network architecture search

Images