Target Image Shapley Processing Across Edge and Cloud
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The calculation of Shapley values in machine learning models is time-consuming and resource-intensive, making it difficult to apply these calculations on terminal devices like edge devices, which are increasingly used for their ease of maintenance and quick response.
Innovation Solution
A method that involves generating a first Shapley value on a terminal device and requesting a second, more accurate Shapley value from a cloud server when needed, utilizing a joint computing architecture between the terminal device and the cloud server to achieve rapid and accurate Shapley value calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Shapley values are calculated using traditional methods, then measurement precision is improved, but loss of time and use of energy increase significantly
Solution Approach 1:
The patent segments the Shapley value calculation into two parts: a first Shapley value calculated locally on the terminal device using an approximation model for quick results, and a second Shapley value calculated on the cloud server using a more accurate method when needed. This segmentation allows the system to achieve both speed and accuracy by dividing the computational task across different platforms with different performance characteristics.
Solution Approach 2:
The patent introduces a cloud server as an intermediary between the terminal device and the accurate Shapley value calculation. The terminal device first attempts to calculate an approximate Shapley value locally, and only invokes the cloud server's more accurate calculation method when the approximation is insufficient. This intermediary approach allows the system to leverage both the speed of local computation and the accuracy of cloud-based computation.
2Measurement precision
If Shapley values are calculated using traditional methods, then measurement precision is improved, but use of energy increases significantly
Solution Approach 1:
The patent segments the energy-consuming calculation task into two parts: lightweight local calculation for quick approximate Shapley values that consume minimal energy, and heavy cloud-based calculation for accurate Shapley values that consume more energy but are only invoked when necessary. This segmentation reduces overall energy consumption by avoiding unnecessary high-cost calculations.
Solution Approach 2:
The patent applies partial action by using the approximation model to calculate Shapley values in most cases where high accuracy is not critical, thereby avoiding the excessive energy consumption of full accurate calculations. The more energy-intensive accurate calculation is performed only partially - when the predetermined condition indicates it is necessary - thus optimizing the balance between accuracy and energy consumption.
3Measurement precision
If accurate Shapley value calculation is performed on terminal devices, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent extracts the complex accurate Shapley value calculation capability from the terminal device and places it on the cloud server. The terminal device only retains the simple approximation model for local calculation, while the complex cloud-based calculation is invoked only when needed. This extraction reduces the complexity burden on the terminal device while maintaining the capability to achieve high accuracy when required.
Solution Approach 2:
The patent introduces the cloud server as an intermediary that handles the complex accurate Shapley value calculation, thereby reducing the complexity requirements for the terminal device. The terminal device simply sends data to the cloud server when accurate calculation is needed, rather than incorporating the complex calculation capability locally.
Data Source
AI summary
Embodiments disclosed herein include a method, an electronic device, and a computer program product for target image processing. The method includes receiving a target image and generating a first Shapley value for a feature of the target image based on the received target image. The method further includes sending, in response to satisfying a predetermined condition, a request for acquiring a second Shapley value to a cloud server. The method further includes receiving the second Shapley value for a latent feature of the target image from the cloud server, where the second Shapley value is more accurate than the first Shapley value. In some embodiments, through joint collaboration between a terminal device such as an edge device and a cloud server, rapid calculation of a Shapley value can be achieved at the terminal device, and accurate calculation of a Shapley value can be achieved at the cloud server.