Image Set Preprocessing Selection for Robust Model Training

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Image sets used for machine learning may contain missing values or outliers, leading to a decrease in the performance of the learning model.

Innovation Solution

A machine learning device that selects an image set similar to the acquired set from a plurality of image sets, performs preprocessing on it, and evaluates multiple learning models based on performance to identify the best model for further training.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If machine learning is performed using an image set that includes images with missing values or outliers, then the machine learning process can proceed without data preprocessing, but the performance of the learning model decreases

Engineering Contradiction:

Improvemachine learning process efficiencyVSAvoidlearning model performance

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system performs preliminary actions by selecting a similar image set in advance and evaluating multiple preprocessing methods on it before actual machine learning. This allows the optimal preprocessing method to be determined beforehand, ensuring high model performance while maintaining efficient training through pre-selected clean data patterns.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system creates a copy of the acquired image set by selecting a similar image set from stored datasets. This similar image set is then used for evaluating preprocessing methods, allowing the system to determine optimal preprocessing without compromising the original training data integrity or requiring extensive preprocessing of the actual training set.

Inventive Principle:

Principle #26Copying

2Reliability

If multiple preprocessing methods are evaluated by training separate learning models, then the optimal preprocessing method can be selected for better performance, but the computational time and resources increase

Engineering Contradiction:

Improvelearning model performanceVSAvoidcomputational time for model evaluation

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system applies partial action by evaluating preprocessing methods on a selected similar image set rather than the entire acquired image set. This partial evaluation using a representative subset allows comparison of multiple preprocessing methods with reduced computational overhead, while still providing sufficient information to determine the optimal preprocessing approach for the actual training data.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS20260017931A1Machine learning device, machine learning method, and computer-readable medium storing machine learning program

Publication Date: 2026.01.15 MITSUBISHI ELECTRIC CORP

AI summary

A machine learning device includes an image set acquiring unit to acquire an image set including images, and an image set selecting unit to select an image set similar to the acquired image set from a plurality of image sets different from the acquired image set. In addition, the machine learning device includes a performance comparison unit, and a preprocessing acquisition unit to select a learning model from a plurality of machine-learned learning models based on a performance comparison result by the performance comparison unit and to acquire preprocessing performed on an image set used for machine learning of the learning model selected. Furthermore, the machine learning device includes a model learning unit to perform the acquired preprocessing on the acquired image set and to cause a learning model that has not yet trained to perform machine learning using the preprocessed image set.