Ensemble Learning Classifier for Semiconductor Defect Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current defect classification approaches in semiconductor manufacturing are time-consuming and limited, requiring manual creation of decision trees with numerous attributes, which affects the quality and efficiency of defect type separability and classifier monitoring.
Innovation Solution
The development of an ensemble learning classifier system that automatically generates defect classifiers based on manual classification and attributes, calculates confidence thresholds, and reports test defects below the threshold for manual review, improving defect classification and separability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual decision trees are created for defect classification, then defect type separability can be achieved, but the process becomes time-consuming and complex
Solution Approach 1:
The system enables automatic defect classification by allowing the defect data and attributes to self-organize into decision trees through automated algorithms. The classifier automatically selects attributes and constructs classification rules without requiring manual intervention, thus achieving defect type separability while eliminating the time-consuming manual tree creation process.
Solution Approach 2:
The patent replaces the manual mechanical process of creating decision trees with an automated computational system. Instead of manually selecting attributes and constructing tree structures, the system uses automated algorithms to generate decision trees from defect data, substituting human manual work with computational automation to achieve both accuracy and efficiency.
2Measurement precision
If manual decision trees with many attributes are created, then defect classification accuracy improves, but device complexity increases
Solution Approach 1:
The system dynamically adjusts the complexity of decision trees based on the specific defect data and classification requirements. Rather than using fixed complex structures, the automated algorithms adaptively select only the necessary attributes and optimize tree structure for each classification task, achieving high accuracy while minimizing unnecessary complexity.
Solution Approach 2:
The patent changes the parameters of the classification system by automatically selecting and weighting different attributes based on their relevance to defect type separation. The system dynamically adjusts attribute importance and tree structure parameters to achieve optimal classification accuracy without requiring predetermined complex structures, thus improving accuracy while controlling complexity.
3Measurement precision
If manual attribute selection is performed for each decision tree node, then classification quality improves, but ease of operation deteriorates
Solution Approach 1:
The system performs automatic attribute selection and classification rule generation, allowing the defect data and algorithms to self-organize the classification structure. This eliminates the need for operators to manually select attributes for each decision tree node, maintaining high classification quality through automated optimization while dramatically improving ease of operation by removing complex manual selection tasks.
4Measurement precision
If current defect classification approaches are used, then defect types can be separated, but productivity decreases due to time-consuming processes
Solution Approach 1:
The patent replaces manual classification processes with automated computational systems that can process large volumes of defect data rapidly. The automated decision tree generation and attribute selection algorithms enable high-throughput classification while maintaining accurate defect type separability, thus improving productivity without sacrificing classification quality.
Solution Approach 2:
The system enables continuous automated classification of defect data without interruption by manual intervention. The automated algorithms can process defect data continuously, generating classification results in real-time, which maintains defect type separability while significantly improving productivity by eliminating bottlenecks associated with manual classification processes.
Data Source
AI summary
Defect classification includes acquiring one or more images of a specimen, receiving a manual classification of one or more training defects based on one or more attributes of the one or more training defects, generating an ensemble learning classifier based on the received manual classification and the attributes of the one or more training defects, generating a confidence threshold for each defect type of the one or more training defects based on a received classification purity requirement, acquiring one or more images including one or more test defects, classifying the one or more test defects with the generated ensemble learning classifier, calculating a confidence level for each of the one or more test defects with the generated ensemble learning classifier and reporting one or more test defects having a confidence level below the generated confidence threshold via the user interface device for manual classification.