Gate Model Selection for Concept-Drift Prediction Precision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge of maintaining prediction precision in machine learning models due to concept drift, where retraining the model results in the loss of past data and models, making it difficult to apply the model effectively.
Innovation Solution
An information processing apparatus that decomposes input data into sub-data sets based on correspondence relations, generates prediction models for each set, and creates a gate model to select the most appropriate model for new input data, enhancing prediction precision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a machine learning model is retrained to adapt to concept drift, then the model can adapt to new data characteristics, but the past data and models are lost resulting in reduced prediction precision
Solution Approach 1:
The patent segments the machine learning model into multiple specialized models, each trained on specific historical data for particular scenarios. Instead of having one model that must be completely retrained, the system maintains multiple models that can be selectively applied. This segmentation allows the system to adapt to concept drift by selecting appropriate models while preserving knowledge from past data in each specialized model.
Solution Approach 2:
The patent performs preliminary actions by pre-training multiple specialized models on different historical datasets before concept drift occurs. These pre-trained models are stored and can be selectively invoked when concept drift is detected, avoiding the need to retrain from scratch and preserving previously learned knowledge.
2Ease of operation
If a single machine learning model is used to maintain simplicity, then the system is easier to manage, but it cannot effectively apply to every situation resulting in reduced prediction precision
Solution Approach 1:
The patent creates a universal model selection mechanism that manages multiple specialized models. The model selection component acts as a universal interface that handles different scenarios by selecting the appropriate specialized model, providing multi-functionality while maintaining ease of operation through automated selection based on input characteristics.
Solution Approach 2:
The patent introduces an intermediary model selection component that mediates between the input data and multiple specialized models. This intermediary automatically determines which model to use based on the characteristics of the input data, simplifying the management complexity by providing a single point of control while still utilizing multiple specialized models for high precision.
Data Source
AI summary
An information processing apparatus of the present disclosure includes: an acquiring unit configured to acquire first data and first models, the first data being composed of pairs of explanatory variables and objective variables classified into a plurality of classifications in accordance with a correspondence relation between the explanatory variable and the objective variable, each of the first models being generated in such a manner as to predict the objective variable from the explanatory variable based on the first data for each of the classifications; and a generating unit configured to generate a second model in accordance with information representing a correspondence relation between the explanatory variable of the first data based on the classification and the first model, the second model for decision making predicting the first model corresponding to the explanatory variable.