Method and system performing pattern clustering
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.
1. A method of clustering patterns of an integrated circuit, the method comprising:
for each pattern among a plurality of patterns,
providing a pattern image to a first model, wherein the first model is trained using a plurality of sample images; and
generating a content latent variable and a pose latent variable using the first model in response to the pattern image, wherein the pose latent variable has a value corresponding to an Euclidean transform of each pattern; and thereafter, grouping a plurality of content latent variables, each respectively corresponding to a pattern among the plurality of patterns, into a plurality of clusters corresponding to the plurality of patterns based on a Euclidean distance;
generating a coordinate in response to the pattern image;
identifying an Euclidean transform of the pattern image by sampling a probability distribution of the pose latent variable;
transforming the coordinate in response to the Euclidean transform of the pattern image to generate a transform coordinate;
providing the transform coordinate to a second model; and
generating a test pattern image using the second model in response to the transform coordinate and the content latent variable.
2. The method of claim 1 , wherein the pose latent variable includes a vector including eight real numbers respectively corresponding to eight transforms included in a dihedral group.
3. The method of claim 1 , further comprising:
providing the content latent variable to the second model, wherein the second model is trained using a plurality of sample content latent variables;
and
training at least one of the first model and the second model using the test pattern image.
4. The method of claim 1 , further comprising:
providing numeric data to the first model, wherein the numeric data represents at least one attribute of a pattern among the plurality of patterns, and
the generating of the content latent variable and the pose latent variable using the first model is performed in response to the pattern image and the numeric data.
5. The method of claim 1 , further comprising:
extracting at least one feature of the pattern image; and
extracting at least one main feature from the at least one feature in response to the content latent variable.
6. The method of claim 5 , wherein the extracting of the at least one main feature includes:
calculating first correlation coefficients between the at least one feature and the plurality of clusters; and
extracting the at least one main feature distinguishing the plurality of clusters in response to the first correlation coefficients.
7. The method of claim 5 , wherein the extracting of the at least one main feature includes:
calculating second correlation coefficients between the at least one feature and content latent variables included in one cluster among the plurality of clusters; and
extracting the at least one main feature distinguishing the content latent variables included in the one cluster in response to the second correlation coefficients.
8. The method of claim 1 , further comprising:
obtaining verification data associated with the pattern image; and
calculating correlation coefficients between the verification data and the plurality of clusters.
9. The method of claim 8 , further comprising:
visualizing the plurality of clusters and the verification data.
10. A method of clustering patterns of an integrated circuit, the method comprising:
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 and a pose 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;
generating a coordinate in response to the pattern image;
identifying an Euclidean transform of the pattern image by sampling a probability distribution of the pose latent variable;
transforming the coordinate in response to the Euclidean transform of the pattern image to generate a transform coordinate;
providing the transform coordinate to a second model; and
generating a test pattern image using the second model in response to the transform coordinate and the content latent variable.
11. The method of claim 10 , further comprising:
extracting a plurality of features in response to at least one of the pattern image and the numeric data; and
extracting at least one main feature from the plurality of features and the plurality of clusters.
12. The method of claim 11 , wherein the extracting of the at least one main feature comprises:
calculating first correlation coefficients between the plurality of features and the plurality of clusters; and
extracting the at least one main feature distinguishing the plurality of clusters in response to the first correlation coefficients.
13. The method of claim 12 , wherein the extracting of the at least one main feature further comprises:
calculating second correlation coefficients between the plurality of features and content latent variables included in one cluster among the plurality of clusters; and
extracting the at least one main feature distinguishing the content latent variables included in the one cluster in response to the second correlation coefficients.
14. The method of claim 11 , further comprising:
visualizing the plurality of clusters and the at least one main feature.
15. A system comprising:
at least one processor; and
a non-transitory storage medium storing instructions, which when executed by the at least one processor, allows the at least one processor to cluster patterns of an integrated circuit,
wherein clustering the patterns comprises:
providing a pattern image corresponding to a first pattern to a first model, wherein the first model is trained by a plurality of sample images;
obtaining a first content latent variable and a first pose latent variable from 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 first pose latent variable represents a Euclidean transform of the first pattern with respect to a representative pattern of a first cluster including the first content latent variable;
generating a first coordinate corresponding to the pattern image;
identifying the Euclidean transform of the first pattern by sampling a probability distribution of the first pose latent variable;
transforming the first coordinate based on the Euclidean transform of the first pattern to generate a transformed first coordinate;
providing the transformed first coordinate to a second model; and
obtaining a first test pattern image from the second model in response to the transformed first coordinate and the first content latent variable.
16. The system of claim 15 , wherein the clustering of patterns further comprises:
providing the first content latent variable to the second model, wherein the second model is trained by a plurality of sample content latent variables;
and
training the first model and the second model based on the pattern image and the first test pattern image.
17. The system of claim 16 , wherein the second model is trained by the plurality of sample content latent variables and a plurality of sample coordinates.
18. The system of claim 15 , wherein the first model is trained using the plurality of sample images and a plurality of sample values, and
the clustering of patterns further comprises providing first numeric data representing at least one attribute of the first pattern to the first model.
19. The system of claim 15 , wherein the clustering of patterns further comprises:
obtaining verification data of the first pattern; and
calculating correlation coefficients between the verification data and the plurality of clusters.