Dual-Weight Sample Training for ML Prediction Precision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Machine learning models, particularly those using supervised learning methods like XGBoost, face challenges in predicting classification results for certain samples, leading to unstable and random predictions due to insufficient features or sparse distribution, resulting in low prediction precision.
Innovation Solution
The method involves maintaining two weights for each sample in the training set, updating them based on predicted losses, and iteratively retraining the model to focus more on incorrectly predicted samples, thereby improving prediction precision and training efficiency through multithreaded processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple classifiers are trained based on a supervised manner, then the machine learning model can predict classification results, but the prediction precision is low for some samples due to difficult-to-predict classification results
Solution Approach 1:
The patent changes the parameter of sample weights by introducing two different weight systems: first weights for correctly predicted samples and second weights for incorrectly predicted samples. This parameter change allows the model to differentiate between easy and hard samples, improving prediction precision for difficult samples while maintaining overall training stability through the dual-weight mechanism.
2Productivity
If the model focuses on all samples equally during training, then training is simpler, but training efficiency is low and prediction precision for critical samples is poor
Solution Approach 1:
The patent applies local quality by assigning different weight qualities to different samples based on their prediction performance. Correctly predicted samples receive first weights while incorrectly predicted samples receive second weights, creating a localized weight distribution that optimizes training efficiency for easy samples while improving prediction precision for hard samples through differentiated attention.
3Device complexity
If the model uses uniform weighting for all samples, then the training process is simpler, but the model cannot adequately attention to incorrectly predicted samples
Solution Approach 1:
The patent segments the sample set into two distinct groups: samples with correct predictions and samples with incorrect predictions. Each segment is assigned a different weight (first weight or second weight), creating a segmented training approach that maintains relative simplicity while significantly improving prediction precision for incorrectly predicted samples through targeted weight differentiation.
Data Source
AI summary
A machine learning model training method includes: training a machine learning model using features of samples in a training set, where a sample in the training set corresponds to an initial first weight and an initial second weight. In one iteration, the method includes: determining a first sample set comprising one or more samples whose corresponding target variables are incorrectly predicted; determining an overall predicted loss of the first sample set based on the predicted losses and corresponding first weights of samples in the first sample set; updating the first weights and second weights of the samples in the first sample set based on the overall predicted loss of the first sample set; and inputting the second weights, the features, and the target variables of the samples in the training set to the machine learning model, and initiating a next iteration of training the machine learning model.