Multimodal Feature Extraction for Substrate Abnormality Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing analysis methods in substrate processing systems are limited to analyzing data from a single modality, failing to consider interrelationships between multiple data types, which hinders comprehensive abnormality detection and factor identification.
Innovation Solution
A method involving a first and second feature value extraction model, along with an abnormality detection model, is used to learn and analyze interrelationships between different data modalities, enabling comprehensive abnormality detection and factor analysis by integrating data from multiple sources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If data analysis is performed using a single modality, then the analysis process is simple, but the comprehensiveness of abnormality detection and factor identification is insufficient
Solution Approach 1:
The patent combines multiple modalities of data (image data, measurement data, log data) into a unified analysis system. The feature value extraction models process different data types simultaneously, and the abnormality detection model integrates features from all modalities to perform comprehensive analysis, thereby reducing information loss while managing system complexity through structured integration.
Solution Approach 2:
The abnormality detection model serves multiple functions by analyzing features from different modalities (image, measurement, log data) simultaneously. The system is designed to handle various data types through a unified architecture, enabling multi-functional analysis without requiring separate systems for each modality.
2Measurement precision
If multiple feature value extraction models are used for different modalities, then the accuracy of abnormality detection is improved, but the complexity of the system increases
Solution Approach 1:
The patent segments the analysis system into multiple specialized feature value extraction models, each dedicated to processing a specific modality (image data, measurement data, log data). This segmentation allows each model to be optimized for its specific data type, improving accuracy while maintaining manageable complexity through modular design.
Solution Approach 2:
The feature values extracted from different modalities serve as intermediaries that bridge the gap between raw multi-modal data and the final abnormality detection. Each feature value extraction model transforms its specific data type into a standardized feature representation that can be integrated by the abnormality detection model, facilitating accurate multi-modal analysis.
Data Source
AI summary
A non-transitory computer-readable medium storing a computer program which, when executed by a computer, causes the computer to execute a process including acquiring a first feature value from a first feature value extraction model, which outputs the first feature value when data of a first modality about substrate processing is received; acquiring a second feature value from a second feature value extraction model which outputs the second feature value when data of a second modality different from the first modality is received; calculating a similarly between the first feature value and second feature value; and training at least one of the first feature value extraction model and the second feature value extraction model based on the similarity.