Ensemble Learning Prediction Method Iterative Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Ensemble learning systems face challenges in effectively training dynamic weighting for predictors, especially in complex sample spaces, leading to inaccurate confidence predictions and difficulty in improving prediction accuracy.
Innovation Solution
The method involves establishing base predictors and initializing sample weights, iteratively evaluating and updating predictor weighting functions and sample weights to select target predictors and weighting functions, combining them into an ensemble predictor, which adapts through multiple iteration rounds to enhance prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dynamic weighting is used in ensemble learning to improve prediction accuracy, then the prediction result improves, but it becomes difficult to effectively train the dynamic weighting when the sample has complicated distribution
Solution Approach 1:
The patent segments the training process into multiple iteration rounds, where in each round only one predictor weighting function is selected and trained based on the current sample weights. This divides the complex task of training all weighting functions simultaneously into manageable sequential steps, making the training process tractable even for complicated sample distributions.
Solution Approach 2:
The patent performs preliminary actions by initializing sample weights and predictor weighting functions before the iterative training process. In each iteration round, the system pre-selects one predictor weighting function to train based on current sample weights, preparing the training process in advance rather than attempting to optimize all weightings simultaneously.
2Ease of manufacture
If fixed weighting is used for base predictors, then the system is simpler to implement, but the prediction result is less accurate compared to dynamic weighting
Solution Approach 1:
The patent transforms the static fixed weighting system into a dynamic system where predictor weighting functions are iteratively trained and updated. The sample weights are dynamically adjusted in each iteration round, allowing the system to adapt to complicated sample distributions and achieve higher prediction accuracy while maintaining a relatively simple implementation through sequential training.
Solution Approach 2:
The system performs self-service by automatically adjusting sample weights and training predictor weighting functions based on prediction performance. The evaluation module assesses prediction results and the system self-corrects by updating weights in subsequent iteration rounds, eliminating the need for complex manual tuning while achieving dynamic weighting benefits.
Data Source
AI summary
An ensemble learning prediction method includes: establishing a plurality of base predictors based on a plurality of training data; initializing a plurality of sample weights of a plurality of sample data and initializing a processing set; in each iteration round, based on the sample data and the sample weights, establishing a plurality of predictor weighting functions of the predictors in the processing set and predicting each of the sample data by each of the predictors in the processing set for identifying a prediction result; evaluating the predictor weighting functions, and selecting a respective target predictor weighting function from the predictor weighting functions established in each iteration round and selecting a target predictor from the predictors in the processing set to update the processing set and to update the sample weights of the sample data.