Integrated Circuit Pattern Clustering With Latent Variables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing complexity of semiconductor processes makes it impractical to thoroughly analyze and verify every pattern in integrated circuits, necessitating a more efficient and accurate method for pattern analysis and verification.
Innovation Solution
A method and system for pattern clustering in integrated circuits using machine learning models to generate latent variables and group patterns based on Euclidean distances, allowing for efficient analysis and verification by clustering patterns with similar attributes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If every pattern in integrated circuits is analyzed and verified individually, then analysis accuracy is improved, but analysis time and computational resources increase significantly
Solution Approach 1:
The patent segments patterns into multiple clusters based on geometric attributes such as area, aspect ratio, and orientation. By dividing the large set of patterns into smaller clusters, the analysis time is reduced while maintaining accuracy through representative pattern selection within each cluster.
Solution Approach 2:
Instead of analyzing every pattern exhaustively, the patent performs partial analysis by selecting representative patterns from each cluster. This partial action approach significantly reduces analysis time while still providing sufficient accuracy for design verification.
2Reliability
If every pattern in integrated circuits is analyzed and verified individually, then analysis completeness is improved, but computational complexity increases significantly
Solution Approach 1:
The patent segments patterns into clusters based on geometric attributes, reducing computational complexity by processing clusters rather than individual patterns. This segmentation maintains analysis completeness through strategic selection of representative patterns from each cluster.
Solution Approach 2:
The patent changes parameters by selecting representative patterns based on geometric attributes (area, aspect ratio, orientation) rather than processing all patterns. This parameter change approach reduces computational complexity while maintaining reliability through attribute-based representation.
3Measurement precision
If patterns are clustered based on multiple geometric attributes, then pattern representation accuracy is improved, but clustering computation time increases
Solution Approach 1:
The patent uses multiple geometric parameters (area, aspect ratio, orientation) to define pattern clusters, improving representation accuracy. The computation time is managed by efficiently calculating these parameters and using them for cluster assignment.
Solution Approach 2:
The patent performs preliminary calculations of geometric attributes (area, aspect ratio, orientation) before clustering. This preliminary action prepares the data in advance, making the actual clustering process more efficient while maintaining high representation accuracy.
Data Source
AI summary
A method of clustering patterns of an integrated circuit includes; providing a pattern image and numeric data, as input data corresponding to a first pattern to a first model, wherein the first model is trained by a plurality of sample images and a plurality of sample values, obtaining a content latent variable using the first model, and grouping a plurality of content latent variables corresponding to a plurality of patterns into a plurality of clusters based on a Euclidean distance, wherein the numeric data represents at least one attribute of the first pattern.