Semiconductor Model Learning Visualization for Process Parameter Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional systems face difficulties in visually and intuitively displaying the learning of machine learning models and statistical models, particularly in semiconductor wafer processing, making it cumbersome to understand the correlations between inputs and outputs, non-linearities, and the extent of input and output spaces.
Innovation Solution
A graphical user interface (GUI) is developed to present model learning, allowing users to visualize high-dimensional input and output spaces, with features like scatter plots and curves that indicate the relationships between input parameters and output features, enabling users to navigate and understand model learning effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If machine learning models and statistical models are used to improve processing procedures in semiconductor chambers, then processing quality and control are improved, but the complexity of understanding and visualizing model learning increases
Solution Approach 1:
The patent introduces an intermediary visualization system that mediates between the complex machine learning models and the user. This system includes a processing device that receives model outputs and generates intuitive visual representations through a graphical user interface, allowing users to understand model learning without directly confronting the mathematical complexity of the models themselves.
Solution Approach 2:
The patent replaces the traditional mechanical approach of directly interacting with complex model parameters and equations with an information-based visualization system. The system substitutes direct mathematical analysis with graphical representations, including plots and visual indicators that convey model learning results in an intuitively understandable format.
2Loss of information
If detailed model outputs are presented to show comprehensive learning results, then information completeness is improved, but ease of operation and interpretation deteriorates
Solution Approach 1:
The patent segments the comprehensive model outputs into distinct visual components through the graphical user interface. The visualization system divides complex data into manageable graphical elements such as plots, curves, and visual indicators that can be independently interpreted. This segmentation allows complete information to be presented while maintaining ease of interpretation through organized, modular visual presentation.
Solution Approach 2:
The patent transforms one-dimensional numerical model outputs into two-dimensional graphical representations. By plotting model learning results on graphical axes and using visual dimensions such as position, size, and shape of visual indicators, the system preserves complete information while making it intuitively interpretable through spatial relationships that are naturally understood by humans.
3Device complexity
If traditional visualization methods are used for model outputs, then system simplicity is maintained, but the ability to detect and measure model learning correlations deteriorates
Solution Approach 1:
The patent employs visual indicators that utilize color variations to represent different aspects of model learning. The graphical user interface incorporates color-coded elements that indicate correlations, relationships, and patterns in the model outputs, making it easier to detect and measure learning results without significantly increasing system complexity.
Solution Approach 2:
The patent uses curved visual representations such as plots and graphical curves to display model learning results. These curved visualizations naturally reveal correlations and relationships in the data that would be difficult to detect in tabular or linear formats, enhancing the ability to measure model learning while maintaining relatively simple system architecture.
Data Source
AI summary
A method includes receiving a first value associated with a first input parameter of a model. The method further includes receiving a first plurality of values associated with a second input parameter of the model. The method further includes providing to the model the first value and the first plurality of values. The method further includes receiving a first plurality of outputs from the model. The method further includes preparing the first plurality of outputs for presentation via a presentation element of a graphical user interface. The presentation element includes two axes. The first axis is associated with a first property of a first feature associated with the output from the model. The second axis is associated with a second property of the first feature.