CART Prediction Models for Application Preloading
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current preloading methods for smart terminal applications are inefficient due to high resource occupation and increased power consumption, leading to reduced terminal fluency and inaccurate prediction of user-initiated applications.
Innovation Solution
A method utilizing Classification and Regression Tree (CART) prediction models to predict and preload target applications based on current state feature information, optimizing resource allocation and improving prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If applications are preloaded in advance to reduce loading time, then application loading speed is improved, but storage space occupation and power consumption increase
Solution Approach 1:
The system performs preliminary actions by preloading applications before they are actually needed. The CART prediction model identifies applications that are likely to be launched soon, and these applications are preloaded into memory in advance, reducing the loading time when users actually access them.
Solution Approach 2:
The system dynamically changes the preloading parameters based on prediction probabilities. Instead of loading all applications or a fixed number, the system adjusts which applications to preload and how many based on the output probabilities from the CART model, optimizing the balance between loading speed and resource consumption.
2Loss of time
If applications are preloaded in advance to reduce loading time, then application loading speed is improved, but storage space occupation increases
Solution Approach 1:
The system performs preliminary actions by preloading applications before they are actually needed. The CART prediction model identifies applications that are likely to be launched soon, and these applications are preloaded into memory in advance, reducing the loading time when users actually access them.
Solution Approach 2:
The system dynamically changes the preloading parameters based on prediction probabilities. Instead of loading all applications or a fixed number, the system adjusts which applications to preload and how many based on the output probabilities from the CART model, optimizing the balance between loading speed and resource consumption.
3Productivity
If traditional preloading methods are used to improve application fluency, then application responsiveness is improved, but prediction accuracy of user-initiated applications deteriorates
Solution Approach 1:
The system replaces traditional mechanical or rule-based preloading methods with a data-driven machine learning approach. The CART prediction model analyzes historical usage patterns and current state features to accurately predict which applications users will launch, substituting guesswork with intelligent prediction.
Solution Approach 2:
The system incorporates feedback mechanisms where the CART model is trained on historical usage data and continuously improves its prediction accuracy. The model learns from past user behavior patterns and adjusts its predictions based on accumulated data, enhancing both accuracy and fluency over time.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
A method for preloading an application, a storage medium, and a terminal device are provided. The method includes the following. Current state feature information of the terminal device is acquired, when an application preloading prediction event is detected to be triggered. The current state feature information is input into a plurality of CART prediction models each corresponding to an application in a preset application set, where each of the CART prediction models is generated based on a usage regularity of an associated application corresponding to historical state feature information of the terminal device. A target application to be initiated is predicted according to output results of the CART prediction models, and then the target application is preloaded.