Artificial intelligence cloud platform

Abstract

The invention discloses an artificial intelligence cloud platform. The intelligent cloud platform is composed of an IaaS framework, a PaaS framework and an SaaS framework in sequence, the cloud platform is constructed in the mode that resources, components and applications are abstracted layer by layer, and the system is completely decoupled, and the IaaS layer distributes and manages bottom hardware resources in a unified mode through the virtualization technology, so that an infrastructure, namely service, of the intelligent cloud platform is constructed; The PaaS layer constructs an autonomous process control engine by performing four-level abstraction on an artificial intelligence application scene; an artificial intelligence component and a system service are constructed; Four-level abstraction is carried out on an artificial intelligence application scene, and the four-level abstraction comprises the following four parts: a data level abstraction KLoud. Cube, a function level abstraction KLoud. Case, a service level abstraction KLoud. Chin and an optimization level abstraction KLoud. Cycle. And the SaaS layer constructs an artificial intelligence application service. Based onan autonomous process control engine, a plurality of flexible and effective operations can be realized, the defects that a stacked open source engine increases the system complexity and is very unfavorable for deployment, debugging and maintenance are solved, and the cloud platform system has high elasticity and high flexibility.

Description

technical field [0001] The invention belongs to the technical field of cloud platforms, and in particular relates to an artificial intelligence cloud platform. Background technique [0002] AI intelligent applications are also developing rapidly with the development of society. At present, AI intelligent applications are mainly divided into traditional single applications, ordinary AI platforms and AI cloud platforms. [0003] Traditional single applications include most of the smart security applications seen on the market and edge computing products launched by Hikvision and Dahua. Such applications basically do not have secondary development capabilities and cannot meet user customization needs. For a single application, equipment damage means that the service is unavailable as a whole, and the maintenance and replacement cycle is long and the cost is high. In addition, a single application is basically inelastic, so its hardware performance must take into account its ...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to solve the problem that the current AI cloud platform uses directly stacked open source big data batch processing / stream processing engines, which increases the complexity of the system, which is very unfavorable for deployment, debugging and maintenance, and the introduction of open source engines reduces the cost of the overall system. issues of elasticity and flexibility, an AI cloud platform is proposed

Images