3.2 Technology Model
Take "cat and dog image recognition", a typical AI industry application practice, as an example, four roles are set and three scenarios are used to describe the specific operation mode of ATMT. Suppose there are four actors (companies or individuals) as follows. A company that specializes in image tagging and expects to provide the data needed for AI to generate revenue.
A company that specializes in developing AI industry applications and expects to earn revenue by relying on the developed AI industry applications. Companies with many graphics servers, with tensorflow and caffe platform runtime environment, expect to make money by renting out these runtime platform resources. An ordinary user, who has a bunch of pictures of cats and dogs in his hands, expects to have tools to help him sort and store them.
Scenario 1: Developers need data to complete machine learning and develop AI industry applications
A: It is the data provider, with pictures of cats and dogs, and the tagging result of each picture (whether it is a cat or a dog).
B: is the resource consumer, the developer of the picture cat and dog identification AI industry application, a company.
Scenario 2: Users use the AI industry application for cat and dog picture recognition on their own computers
B: AI industry application provider, deploying the developed cat and dog recognition AI industry application information to the blockchain and providing the download of AI industry application executable file data.
D: Ordinary users, who have their own graphics card computers, can support running cat and dog identification AI industry applications.
Scenario 3: Users use the cat and dog picture recognition AI industry application on the third party's platform The role is set as above, and the specific process of three-party interaction is as follows.
B deploys the developed cat and dog recognition AI industry application information to the blockchain and provides the download of AI industry application executable file data.
User D does not have a suitable computer to run the cat and dog identification AI industry application and needs to borrow the platform provided by C. The address of the running resource ADU provided by C needs to be transferred to C.
C allocates a graphics server with AI industry application running platform resources to D.
D pays ATMT to B, the address of the ADU unit of the cat and dog identification application provided by B.
D initiates loading and running the dog and cat identification AI industry application provided by B on the AI industry application runtime platform allocated by C. The image can then be uploaded and the classification completed.
Details.
The data provider, the application provider, and the running platform resource provider all deploy their resources, by creating a transaction, to the ATMT blockchain.
The resource consuming party needs to transfer ATMT to the address of the ADU unit of the said resource to apply for the right to use it.
In the subsequent usage process, the resource consumer is required to provide the public key used in transferring the ATMT to: the data provider, the application provider, and the runtime platform resource provider. This way, the resource provider has enough information to verify the identity of the user and whether sufficient ATMT has been paid. All applications are agreed to be developed and run under the DockerIMG runtime environment released by the ATMT platform to ensure compatibility.
Last updated
