Iterative Machine Learning Model Feedback Loop for Estimation Accuracy

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Machine learning methods for predicting an objective variable from an explanatory variable face a plateau in accuracy.

Innovation Solution

A data estimation technique that creates a machine learning model to estimate an objective variable, adds the estimated objective variable back into the explanatory variable group, and iteratively builds subsequent models to enhance prediction accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional machine learning methods are used to predict objective variables from explanatory variables, then the model can be trained and deployed, but the accuracy in objective variable estimation reaches a plateau and cannot be further improved

Engineering Contradiction:
Improveaccuracy in objective variable estimationVSAvoidmodel training complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies feedback by taking the estimated objective variable from one model and feeding it back as an explanatory variable for training the next model. This creates a iterative loop where each model's output becomes the next model's input, continuously improving estimation accuracy by incorporating previously derived information into subsequent training processes.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent performs preliminary actions by first training a model to estimate objective variables before using those estimates as explanatory variables for subsequent models. This preparatory step ensures that each iterative cycle has the necessary preliminary information (estimated objective variables) ready to enhance the next model's training, breaking the accuracy plateau through structured preliminary processing.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If multiple machine learning models are trained iteratively with added objective variables, then estimation accuracy improves, but the training process becomes more complex and time-consuming

Engineering Contradiction:
Improveprediction accuracyVSAvoidtraining time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent maintains continuity of useful action by making each model's output directly useful for the next model's input without interruption or waste. The estimated objective variables are continuously generated and immediately utilized as explanatory variables in the next iteration, ensuring that each training cycle builds directly on the previous one without loss of useful information or unnecessary delays.

Inventive Principle:
Principle #20Continuity of useful action

Data Source

PatentEP4120147B1Data estimation device, method, and program
Publication Date: 2024.06.26 ASICS CORP
  • EP4120147B1 patent drawingFigure 1
  • EP4120147B1 patent drawingFigure 2
  • EP4120147B1 patent drawingFigure 3

AI summary

Provided is a data estimation technique capable of improving accuracy in estimation of an objective variable from an explanatory variable. In a data estimation device (100), a learning unit (20) creates, using training data including an explanatory variable and an objective variable, a machine learning model that estimates an objective variable from an explanatory variable. The learning unit (20) creates a machine learning model Mi that estimates an objective variable Oi from an explanatory variable group Ei including one or more explanatory variables, sets a new explanatory variable group Ei + 1 by adding the objective variable Oi estimated by the machine learning model Mi to the explanatory variable group Ei, and creates a machine learning model Mi + 1 that estimates an objective variable Oi + 1 from the explanatory variable group Ei + 1 (where i = 1). The learning unit (20) repeatedly creates a machine learning model while i is in a range of from 2 to (n - 1) (n is a natural number greater than or equal to 2).