Model parameter training method, device and system

Abstract

The embodiment of the invention discloses a model parameter training method, a device and a system, and the method, the device and the system are used for rapidly carrying out image retrieval or parameter training of image classification. The method comprises the steps of using model parameters to carry out iterative computation on an objective function, wherein the objective function is a cost function used for image training; if the result of the iterative computation does not meet the termination condition, determining the first gradient of the objective function on the model parameters, and updating the learning rate according to the parameter distribution characteristics of the model parameters in the objective function; updating the model parameters according to the learning rate and the first gradient; repeating the previous steps until the result of the iterative computation meets the termination condition; obtaining the model parameter corresponding to the result of the iterative computation meeting the termination condition.

Description

technical field [0001] The present invention relates to communication technology, in particular to a model parameter training method, device and system. Background technique [0002] There is a semantic gap problem in the traditional method of searching images based on keywords, which often leads to users not being able to retrieve the images they want to search. The content-based image retrieval (CBIR, Content Based Image Retrieval) method is a retrieval method that is more similar to human thinking. The current CBIR system mainly relies on some shallow machine learning algorithms, and its performance is greatly restricted. Deep learning is the most eye-catching direction in the field of machine learning in recent years. Its motivation is to establish and simulate the neural network of human brain for analysis and learning, which imitates the mechanism of human brain to interpret data, such as images, sounds and texts. The concept of deep learning originates from the res...

AI Technical Summary

Problems solved by technology

[0011] The disadvantage of the stochastic gradient descent method in the prior art is that manual parameter selection is required, including learning rate, termination conditions, etc.

When the learning rate is set too small, the training process will be very slow; when the learning rate is set too large, it may skip the local optimal solution when updating the model parameters for iterative calculation, so that the convergence speed will not decrease but increase, or even lead to non-convergence

Images