Intelligent model construction method based on meta-features

Abstract

The invention discloses an intelligent model construction method based on meta-features. An intelligent model description method based on meta-features is adopted, so various key features of the intelligent model are described and abstracted; parameter values in the established model description file are set, thus implementing reconstruction of intelligent models; under this architecture, developers do not need to understand complex knowledge backgrounds; instead, intelligent model design and development meeting different business requirements can be completed only by understanding the model description file and configuring the proper meta-feature state, so that the development efficiency of the intelligent application is effectively improved, the learning cost in development is reduced, and meanwhile, the selectable range of the intelligent model in the intelligent application development is expanded.

Description

technical field [0001] The invention belongs to the technical field of machine learning, and in particular relates to an intelligent model construction method based on meta-features. Background technique [0002] At present, artificial intelligence technology is widely used in various fields of life. The key to the application of artificial intelligence technology is to design and develop machine learning models. Most of the existing technical methods used to support the development of machine learning models define a set of design frameworks, and developers call the interfaces in the frameworks to realize the internal calculations of the models. Mainstream design frameworks have Caffe, Tensorflow, Pytorch and Theano, and the closest to the present invention is the Caffe design framework. Caffe adopts the method of separating representation and implementation to realize the design and operation of the deep neural network model. It uses Protocol Buffer to define the model f...

AI Technical Summary

Problems solved by technology

[0003] When using the existing technology for intelligent model design, there will be many problems in the design framework interface, complex interface parameters, and a high starting point for learning the framework itself.

Usually, in order to realize the intelligent model of a certain function, it is necessary to master a complete set of design framework, which leads to high development cost of intelligent model, high learning threshold, and low development efficiency, which makes ordinary program developers discouraged from machine learning technology.

Images