Task processing method and device based on machine learning, and terminal device

Abstract

The invention provides a task processing method and device based on machine learning, and a terminal device, wherein the method comprises the steps of: adding task processing requests into corresponding task processing queues according to the types of the task processing requests of a machine learning model, wherein the task processing queues include a combined request queue and an ordered requestqueue; the combined request queue includes parameter updating requests; and the ordered request queue includes parameter obtaining requests; when a combination condition is satisfied, combining the various parameter updating requests in the combined request queue, and updating model parameters of the machine learning model according to the combined parameter updating requests; and, according to the sorting sequence of the parameter obtaining requests in the ordered request sequence, sequentially obtaining the model parameters of the machine learning model, and returning to a corresponding client side. By adoption of the task processing method and device based on machine learning, and the terminal device in the invention, the task processing requests of machine learning can be processed ina classified manner; mutual dependence among different types of requests can be avoided; and the task processing efficiency in a machine learning process is increased.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a task processing method, device and terminal equipment based on machine learning. Background technique [0002] Machine learning (Machine Learning, ML) is mainly the process of adjusting its own model structure or model parameters by learning the features in the training data. Among them, machine learning will train and update a large number of parameters in the model during the training process. The training data may include data in text format. One line of these texts is a training sample, the total number of entries (that is, the number of lines) of the training sample will reach the order of trillions, and the features in the sample will also reach the order of hundreds of billions. As the number of features increases, the speed of the machine learning training process also becomes slower. [0003] like figure 1 As shown, the traditional machine learning traini...

AI Technical Summary

Problems solved by technology

[0005] Embodiments of the present invention provide a machine learning-based task processing method, device, storage medium, and terminal equipment to solve or alleviate one or more of the above technical problems in the prior art

Images