AI Application Version Compression for Real-Time Terminal Updates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for updating AI applications on terminal equipment are inefficient due to limitations in storage capacity and computing power, leading to long development cycles and performance degradation.
Innovation Solution
A method involving real-time monitoring of input data and operation states to determine when updates are needed, with a full version of the AI application being updated on server equipment and compressed to a lite version for deployment on terminal equipment, allowing for rapid updates and improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If AI application is updated in real-time on terminal equipment, then accuracy and performance are improved, but storage capacity and computing power limitations are exceeded
Solution Approach 1:
The patent divides the AI application into two versions: a full version deployed on server equipment and a lite version deployed on terminal equipment. The full version contains complete model parameters and structures for high accuracy, while the lite version contains only essential parameters for basic functionality. This segmentation allows the terminal to operate with limited storage while maintaining access to high-accuracy processing through the server.
Solution Approach 2:
The patent introduces a version management module that acts as an intermediary between the server and terminal equipment. This module coordinates the deployment of different AI application versions, manages parameter synchronization, and handles the transition between lite and full versions. The intermediary enables real-time updates without requiring the terminal to store complete model data.
2Reliability
If AI application is updated in real-time on terminal equipment, then accuracy and performance are improved, but development cycle becomes longer
Solution Approach 1:
The patent implements preliminary action by pre-training the full version of the AI application on the server equipment before deploying to terminals. The version management module pre-configures the lite version with essential parameters and establishes the update mechanism in advance. This allows the terminal to quickly switch to updated versions without experiencing lengthy development or deployment cycles.
Solution Approach 2:
The patent ensures continuity of useful action by maintaining both lite and full versions simultaneously and enabling seamless transitions between them. The version management module continuously monitors performance and automatically triggers updates without interrupting service. This continuous update mechanism eliminates development cycle delays by keeping the AI application constantly improved without stopping operations.
3Measurement precision
If full version of AI application is deployed on terminal equipment, then processing accuracy is improved, but device complexity increases
Solution Approach 1:
The patent applies local quality by giving different versions of the AI application different functional characteristics suited to their deployment environments. The lite version on terminals has simplified parameters optimized for low-resource operation, while the full version on servers has complete parameters for high-accuracy processing. Each version has quality tailored to its specific location and capabilities.
Solution Approach 2:
The patent introduces dynamics by making the AI application configuration flexible and adaptable. The version management module dynamically selects which version to use based on available resources, performance requirements, and update status. This dynamic approach allows the system to adjust between simplified and complex configurations without requiring permanent commitment to one complexity level.
Data Source
AI summary
The present disclosure relates to a method for managing an artificial intelligence (AI) application, a device, and a program product. One method comprises receiving input data to be processed by the AI application; updating a first version of the AI application with the input data to generate a second version, wherein the first version is deployed at server equipment; compressing the second version of the AI application to generate a third version of the AI application; and deploying the third version of the AI application to terminal equipment to replace a fourth version of the AI application deployed at the terminal equipment, wherein the fourth version of the AI application is used for processing the input data received at the terminal equipment. A device and a computer program product corresponding thereto are provided.