Generative adversarial networks (GANs) for simulating specimen images

1. A system configured to generate a simulated image of a specimen, comprising:

one or more computer subsystems; and

one or more components executed by the one or more computer subsystems, wherein the one or more components comprise a generative adversarial network trained with a training set comprising portions of design data for one or more specimens designated as training inputs and corresponding images of the one or more specimens designated as training outputs; and

wherein the one or more computer subsystems are configured for:

generating a portion of design data for a specimen input to the generative adversarial network by modifying original design data for the specimen with a synthetic defect; and

generating a simulated image for the specimen by inputting the portion of the design data for the specimen into the generative adversarial network, wherein the simulated image comprises an augmented defect image of the synthetic defect.

2. The system of claim 1 , wherein the generative adversarial network is configured as a conditional generative adversarial network.

3. The system of claim 1 , wherein a number of actual defects detectable on the one or more specimens and the specimen by an inspection system is insufficient for training a machine learning model.

4. The system of claim 1 , wherein the one or more computer subsystems are further configured for determining one or more characteristics of the synthetic defect based on one or more defects detected on the one or more specimens.

5. The system of claim 1 , wherein the one or more computer subsystems are further configured for determining one or more characteristics of the synthetic defect without information for one or more actual defects detected on the one or more specimens or the specimen.

6. The system of claim 1 , wherein the synthetic defect is a first type of defect of interest, wherein the one or more computer subsystems are further configured for generating an additional portion of the design data for the specimen by modifying the original design data for the specimen with an additional synthetic defect, wherein the additional synthetic defect is a second type of defect of interest different than the first type, and wherein the one or more computer subsystems are further configured for generating an additional simulated image for the specimen by inputting the additional portion of the design data into the generative adversarial network.

7. The system of claim 1 , wherein the one or more computer subsystems are further configured for determining one or more characteristics of the synthetic defect based on one or more defects detected on one or more additional specimens, wherein the one or more additional specimens are formed in one or more process steps, wherein a change in the one or more process steps occurs prior to the one or more process steps being used to form the specimen, wherein a machine learning model was trained to perform one or more functions for the one or more additional specimens, and wherein the one or more computer subsystems are further configured to re-train the machine learning model using the simulated image.

8. The system of claim 1 , wherein the corresponding images of the one or more specimens are generated by a first mode of an imaging system, wherein the one or more components further comprise an additional generative adversarial network trained with an additional training set comprising the portions of the design data for the one or more specimens designated as additional training inputs and corresponding additional images of the one or more specimens designated as additional training outputs, wherein the corresponding additional images are generated by a second mode of the imaging system different than the first mode, and wherein the one or more computer subsystems are further configured for generating an additional simulated image for the specimen by inputting the portion of the design data for the specimen into the additional generative adversarial network.

9. The system of claim 1 , wherein the one or more specimens used to generate the training set are different from the specimen for which the generative adversarial network generates the simulated image.

10. The system of claim 1 , wherein the one or more specimens used to generate the training set are different from the specimen for which the generative adversarial network generates the simulated image, and wherein a design for the specimen for which the simulated image is generated is different than one or more designs for the one or more specimens.

11. The system of claim 1 , wherein the one or more computer subsystems are further configured to train a machine learning model using the simulated image.

12. The system of claim 11 , wherein the machine learning model is configured for nuisance filtering.

13. The system of claim 11 , wherein the machine learning model is configured for defect classification.

14. The system of claim 1 , wherein the one or more computer subsystems are further configured for generating an additional simulated image for the specimen by inputting a portion of the original design data for the specimen into the generative adversarial network, wherein the additional simulated image is a reference image, and wherein the system further comprises an inspection system configured for detecting defects on the specimen by subtracting the reference image from images of the specimen generated by the inspection system.

15. The system of claim 1 , wherein the simulated image is a simulated optical image.

16. The system of claim 1 , wherein the specimen is a wafer.

17. A non-transitory computer-readable medium, storing program instructions executable on one or more computer systems for performing a computer-implemented method for generating a simulated image of a specimen, wherein the computer-implemented method comprises:

generating a portion of design data for a specimen input to a generative adversarial network by modifying original design data for the specimen with a synthetic defect;

generating a simulated image for the specimen by inputting the portion of the design data for the specimen into the generative adversarial network, wherein the simulated image comprises an augmented defect image of the synthetic defect, and wherein said inputting is performed by the one or more computer systems;

wherein one or more components are executed by the one or more computer systems, wherein the one or more components comprise the generative adversarial network; and

wherein the generative adversarial network is trained with a training set comprising portions of design data for one or more specimens designated as training inputs and corresponding images of the one or more specimens designated as training outputs.

18. A computer-implemented method for generating a simulated image of a specimen, comprising:

generating a portion of design data for a specimen input to a generative adversarial network by modifying original design data for the specimen with a synthetic defect;

generating a simulated image for the specimen by inputting the portion of the design data for the specimen into the generative adversarial network, wherein the simulated image comprises an augmented defect image of the synthetic defect, and wherein said inputting is performed by one or more computer subsystems;

wherein one or more components are executed by the one or more computer subsystems, wherein the one or more components comprise the generative adversarial network; and

wherein the generative adversarial network is trained with a training set comprising portions of design data for one or more specimens designated as training inputs and corresponding images of the one or more specimens designated as training outputs.