Algorithm Download via Segmented SDK and Cloud Version Codes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods require regenerating a software development kit (SDK) each time an artificial intelligence algorithm is downloaded, making it difficult for clients to obtain algorithms efficiently.
Innovation Solution
A method where a client downloads a corresponding version code from a cloud based on an algorithm identifier and capability description, eliminating the need for SDK regeneration, and allowing easy access to intelligent capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a software development kit (SDK) is used to provide artificial intelligence capability to a client, then the client can obtain the algorithm, but a software development kit needs to be regenerated each time the client obtains an algorithm, making it difficult for the client to obtain the algorithm
Solution Approach 1:
The patent segments the SDK into two independent parts: an algorithm identifier (small data) and the actual algorithm (large data). The client first downloads the small algorithm identifier from the server, which contains a pointer to the algorithm location. Then the client downloads only the specific algorithm file from the specified location, rather than downloading the entire SDK again. This segmentation allows efficient algorithm updates without requiring full SDK regeneration.
Solution Approach 2:
The patent extracts the essential information (algorithm identifier containing location pointer) from the complete SDK and separates it from the full algorithm package. The server stores only the compact algorithm identifier in the SDK, while the complete algorithm files are stored separately. This extraction enables clients to obtain algorithms efficiently by downloading only the necessary components rather than regenerating entire SDKs.
2Reliability
If the entire software development kit is downloaded to ensure complete algorithm functionality, then all capabilities are available, but the download time and data consumption increase significantly
Solution Approach 1:
The patent divides the SDK download process into two stages: first downloading a small algorithm identifier (quick download), then downloading only the specific algorithm file needed (controlled download). This ensures complete functionality is achieved while minimizing initial download time and allowing users to control the second download based on their specific needs and constraints.
Solution Approach 2:
The server performs preliminary action by pre-storing algorithm identifiers with location pointers in the SDK. This preparation allows clients to quickly obtain the identifier and then efficiently locate and download the specific algorithm file without needing to process or analyze the entire SDK package, significantly reducing download time while maintaining functionality completeness.
Data Source
Figure 1
Figure 2~3
Figure 4~5
AI summary
This application discloses an algorithm download method, a device, and a related product. The method includes: obtaining an algorithm identifier of an algorithm and a capability description of a client (S103); sending the algorithm identifier and the capability description to a cloud (S104); and receiving a version code that is of the algorithm and that is returned by the cloud (S106), where the version code is obtained by the cloud by searching based on the algorithm identifier and the capability description. According to the foregoing solution, an algorithm can be easily downloaded.