Neural network architecture evaluation method based on attribute graph optimization

Abstract

The invention discloses a neural network architecture evaluation method based on attribute graph optimization, and the method comprises the steps: modeling a neural network architecture as an attribute graph, and constructing a Bayesian graph neural network agent model; randomly generating, training, and testing a group of neural network architectures, taking the group of neural network architectures and performance indexes corresponding to testing as an initial training set, wherein the training set is used for training a Bayesian graph neural network agent model; according to the current training set, generating a new neural network candidate set through an evolutionary algorithm and training a Bayesian graph neural network agent model; selecting a potential individual from the neural network candidate set by maximizing a collection function, then training and testing the individual, and adding the individual and a performance index corresponding to the test into the current training set; and under the constraint of fixed cost, repeating the above steps until the best neural network architecture and the weight corresponding to the architecture are obtained in the current training set. Compared with the prior art, the method has the advantage that the model with a better effect than manual design can be quickly found.

Description

technical field [0001] The invention relates to the field of automatic machine learning, in particular to a neural network architecture evaluation method based on attribute graph optimization. Background technique [0002] Deep learning has been successfully applied to many fields, such as image recognition, speech recognition, machine translation, etc. Under normal circumstances, these deep learning models need to be carefully designed by excellent experts. Due to the huge search space, on the one hand, it is very time-consuming to actually train a neural network, and on the other hand, designing these models requires a lot of energy. To solve this problem, we use a Bayesian optimization method. The conventional approach to automatic machine learning problems is to formalize the machine learning process as a black-box optimization task, and Bayesian optimization has been well applied in automatic machine learning. Bayesian optimization can be applied to the neural network...

AI Technical Summary

Problems solved by technology

However, only simple fusion of these architecture information is used to compare the similarity of the two networks, and the architectural features of the neural network are not fully searched. The characteristics of these networks need to be manually analyzed and designed, and the similarity of the design architecture is very subjective. Factors, different similarity functions have a great influence on the final result

Images