Pattern grouping method based on machine learning
A pattern grouping method may include receiving an image of a first pattern, generating a first fixed-dimensional feature vector using trained model parameters applying to the received image, and assigning the first fixed-dimensional feature vector a first bucket ID. The method may further include creating a new bucket ID for the first fixed-dimensional feature vector in response to determining that the first pattern does not belong to one of a plurality of buckets corresponding to defect patterns, or mapping the first fixed-dimensional feature vector to the first bucket ID in response to determining that the first pattern belongs to one of a plurality of buckets corresponding to defect patterns.
1. A method for group mapping, comprising:
receiving an image of a first pattern;
generating a first fixed-dimensional feature vector using trained model parameters, the model parameters being based on the received image; and
assigning the first fixed-dimensional feature vector a first bucket identity (ID).
2. The method of claim 1 , wherein assigning the first fixed-dimensional feature vector the first bucket ID further comprises:
creating a new bucket ID for the first fixed-dimensional feature vector in response to a determination that the first pattern does not belong to one of a plurality of buckets corresponding to defect patterns.
3. The method of claim 1 , wherein assigning the first fixed-dimensional feature vector the first bucket ID further comprises:
mapping the first fixed-dimensional feature vector to the first bucket ID in response to a determination that the first pattern belongs to one of a plurality of buckets corresponding to defect patterns.
4. The method of claim 2 , wherein the defect patterns comprise GDS information associated with defects.
5. The method of claim 4 , wherein the defect patterns comprise information derived from the GDS information that includes number of sides, number of angles, dimension, shape, or a combination thereof.
6. The method of claim 1 , wherein the fixed-dimensional feature vector is a one-dimensional feature vector.
7. The method of claim 1 , wherein the trained model parameters are obtained by:
obtaining a plurality of images of a plurality of patterns with assigned bucket IDs; and
training model parameters for a deep learning network.
8. The method of claim 7 , wherein the trained model parameters are further obtained by:
applying parameters of a single polygon located in a center of one of a plurality of images for the deep learning network.
9. The method of claim 1 , further comprising:
pre-training a linear classifier network based on Graphic Data System (GDS) of a sample.
10. The method of claim 9 , wherein the pre-training of the linear classifier network includes:
identifying a portion of the GDS that is associated with a region,
generating label data for the portion of the GDS that indicates a location of the region, and that indicates a type of shape of polygon data associated with the portion of the GDS, and
pre-training the linear classifier network based on the portion of the GDS and based on the label data.
11. A system for pattern grouping, comprising:
a charged particle beam apparatus including a detector;
an image acquirer that includes circuitry to receive a detection signal from the detector and construct an image including a first feature; and
a controller with at least one processor and a non-transitory computer readable medium comprising instructions that, when executed by the processor, cause the system to:
determine pattern data corresponding to the first feature;
determine, by a classifier, a feature vector corresponding to the first feature; and
determine a pattern group of the first feature based on the feature vector.
12. The system of claim 11 , wherein the pattern data comprises a sample design plan.
13. The system of claim 11 , wherein the classifier is a linear classifier and the instructions further cause the system to:
train the linear classifier using a plurality of defect patterns having a plurality of vectors corresponding to defect features.
14. The system of claim 13 , wherein the plurality of defect patterns is retrieved from a storage.
15. A non-transitory computer readable medium storing a set of instructions that is executable by one or more processors of a system to cause the system to perform a method comprising:
acquiring an image of a sample;
identifying a first feature in the image;
determining pattern data corresponding to the first feature;
determining, by a linear classifier, a feature vector corresponding to the first feature; and
determining a pattern group of the first feature based on the feature vector.
16. The method of claim 1 , further comprising:
generating a plurality of fixed-dimensional feature vectors that represent the received image; and
for each of a plurality of patterns included in the received image, assigning a bucket ID based on the plurality of fixed-dimensional feature vectors.
17. The system of claim 11 , wherein the non-transitory computer readable medium comprises instructions that, when executed by the processor, cause the system to:
generate a plurality of fixed-dimensional feature vectors that represent the received image; and
for each of a plurality of patterns included in the received image, assign a bucket ID based on the plurality of fixed-dimensional feature vectors.
18. The non-transitory computer readable medium of claim 15 , wherein the set of instructions are executable by one or more processors of a system to cause the system to perform:
generating a plurality of fixed-dimensional feature vectors that represent the received image; and
for each of a plurality of patterns included in the received image, assigning a bucket ID based on the plurality of fixed-dimensional feature vectors.