Accelerated training of a machine learning based model for semiconductor applications
Methods and systems for accelerated training of a machine learning based model for semiconductor applications are provided. One method for training a machine learning based model includes acquiring information for non-nominal instances of specimen(s) on which a process is performed. The machine learning based model is configured for performing simulation(s) for the specimens. The machine learning based model is trained with only information for nominal instances of additional specimen(s). The method also includes re-training the machine learning based model with the information for the non-nominal instances of the specimen(s) thereby performing transfer learning of the information for the non-nominal instances of the specimen(s) to the machine learning based model.
1. A system configured to train a machine learning based model, comprising:
one or more computer subsystems; and
one or more components executed by the one or more computer subsystems and stored on a non-transitory computer-readable medium, wherein the one or more components comprise:
a machine learning based model configured for performing one or more simulations for specimens, wherein the machine learning based model is initially trained with only information for nominal instances of one or more of the specimens;
wherein the one or more computer subsystems are configured for:
acquiring information for non-nominal instances of one or more of the specimens on which a process is performed; and
re-training the machine learning based model with the information for the non-nominal instances of the one or more of the specimens thereby performing transfer learning of the information for the non-nominal instances of the one or more of the specimens to the machine learning based model.
2. The system of claim 1 , wherein performing the one or more simulations comprises generating one or more simulated images for one of the specimens, and wherein the one or more simulated images illustrate how the one of the specimens appears in one or more actual images of the one of the specimens generated by an imaging system.
3. The system of claim 2 , wherein the imaging system is an optical based imaging system.
4. The system of claim 2 , wherein the imaging system electron beam based imaging system.
5. The system of claim 1 , wherein performing the one or more simulations comprises generating one or more simulated measurements for one of the specimens, and wherein the one or more simulated measurements represent output generated for the one of the specimens by a metrology system.
6. The system of claim 1 , wherein the non-nominal instances comprise instances of defects on the one or more specimens.
7. The system of claim 1 , wherein the non-nominal instances comprise instances of defects on the one or more specimens, and wherein the one or more specimens comprise one or more actual specimens on which the process is performed with two or more different values of one or more parameters of the process.
8. The system of claim 7 , wherein the process is performed with the two or more different values of the one or more parameters of the process in a process window qualification method.
9. The system of claim 7 , wherein the process is performed with the two or more different values of the one or more parameters of the process in a process window qualification method designed for overlay margin determination.
10. The system of claim 7 , wherein the process is performed with the two or more different values of the one or more parameters of the process in a focus exposure matrix method.
11. The system of claim 1 , wherein the acquired information is generated from synthetic design data for the one or more specimens produced by an electronic design automation tool.
12. The system of claim 1 , wherein the non-nominal instances comprise instances of defects on the one or more specimens, and wherein the defects comprise one or more synthetic defects generated by altering a design for the one or more specimens to create the synthetic defects in the design.
13. The system of claim 12 , wherein the one or more components further comprise an inception module configured for altering the design to create the synthetic defects in the design.
14. The system of claim 1 , wherein the non-nominal instances comprise instances of defects on the one or more specimens, wherein the defects comprise one or more synthetic detects generated by altering a design for the one or more specimens to create the synthetic defects in the design, and wherein the information for the non-nominal instances comprises output generated by an imaging or metrology system for the one or more specimens on which the synthetic defects are printed.
15. The system of claim 1 , wherein the non-nominal instances comprise instances of defects on the one or more specimens, wherein the defects comprise one or more synthetic defects generated by altering a design for the one or more specimens to create the synthetic defects in the design, wherein the information for the non-nominal instances comprises output of another model, and wherein the output of the other model represents the one or more specimens on which the synthetic defects are printed.
16. The system of claim 1 , wherein the non-nominal instances comprise instances of defects on the one or more specimens, wherein the defects comprise one or more synthetic defects generated by altering a design for the one or more specimens to create the synthetic defects in the design, wherein the information for the non-nominal instances comprises output of another model, and wherein the output of the other model illustrates how the one or more specimens on which the synthetic defects are printed appear in one or more actual images of the one or more specimens generated by an imaging system.
17. The system of claim 1 , wherein the non-nominal instances comprise instances of defects on the one or more specimens, wherein the defects comprise one or more synthetic defects generated by altering a design for the one or more specimens to create the synthetic defects in the design, wherein the information for the non-nominal instances comprises output of another model, and wherein the output of the other model represents output generated by a metrology system for the one or more specimens on which the synthetic defects are printed.
18. The system of claim 1 , wherein the non-nominal instances comprise instances of defects on the one or more specimens, wherein the defects comprise one or more synthetic defects generated by altering a design for the one or more specimens to create the synthetic defects in the design, wherein the information for the non-nominal instances comprises output of another model, wherein the output of the other model represents output generated by another system for the one or more specimens on which the synthetic defects are printed, and wherein the other model is a deep generative model.
19. The system of claim 1 , wherein the non-nominal instances comprise instances of defects on the one or more specimens, wherein the defects comprise one or more synthetic defects generated by altering a design for the one or more specimens to create the synthetic defects in the design, and wherein the information for the non-nominal instances comprises the altered design.
20. The system of claim 1 , wherein the one or more components further comprise a deep generative model configured to create the information for the nominal instances of the one or more specimens.
21. The system of claim 1 , wherein the nominal instances of the one or more specimens comprise natural scene images.
22. The system of claim 1 , wherein the nominal instances of the one or more specimens comprise more than one type of data.
23. The system of claim 1 , wherein the machine learning based model is a discriminative model.
24. The system of claim 1 , wherein the machine learning based model is a neural network.
25. The system of claim 1 , wherein the machine learning based model is a convolution and deconvolution neural network.
26. The system of claim 1 , wherein the one or more components further comprise one or more additional components, wherein the re-training is performed using the one or more additional components, and wherein the one or more additional components comprise a common mother network for all layers on the specimens, a grand common mother network for all layers on the specimens, an adversarial network, a deep adversarial generative network, an adversarial autoencoder, a Bayesian Neural Network, a component configured for a variational Bayesian method, a ladder network, or some combination thereof.
27. The system of claim 1 , wherein the re-training comprises transferring all weights of convolutional layers of the initially trained machine learning based method and fine tuning weights of fully connected layers of the initially trained machine learning based method.
28. The system of claim 1 , further comprising an electron beam based imaging subsystem configured to generate electron beam images of the specimens, wherein the one or more computer subsystems are further configured for receiving the electron beam images from the electron beam based imaging subsystem.
29. The system of claim 1 , further comprising an optical based imaging subsystem configured to generate optical images of the specimens, wherein the one or more computer subsystems are further configured for receiving the optical images from the optical based imaging subsystem.
30. The system of claim 1 , further comprising an inspection subsystem configured to generate output for the specimens, wherein the one or more computer subsystems are further configured for receiving the output from the inspection subsystem and detecting defects on the specimens based on the output.
31. The system of claim 1 , further comprising a defect review subsystem configured to generate output for defects detected on the specimens, wherein the one or more computer subsystems are further configured for receiving the output from the defect review subsystem and determining properties of the defects detected on the specimens based on the output.
32. The system of claim 1 , further comprising a metrology subsystem configured to generate output for the specimens, wherein the one or more computer subsystems are further configured for receiving the output from the metrology subsystem and determining properties of the specimens based on the output.
33. The system of claim 1 , further comprising a semiconductor fabrication subsystem configured to perform one or more fabrication processes on the specimens.
34. The system of claim 1 , wherein the specimens comprise wafers.
35. The system of claim 1 , wherein the specimens comprise reticles.
36. A non-transitory computer-readable medium, storing program instructions executable on one or more computer systems for performing a computer-implemented method for training a machine learning based model, wherein the computer-implemented method comprises:
acquiring information for non-nominal instances of one or more specimens on which a process is performed, wherein a machine learning based model is configured for performing one or more simulations for the specimens, and wherein the machine learning based model is initially trained with only information for nominal instances of one or more additional specimens; and
re-training the machine learning based model with the information for the non-nominal instances of the one or more specimens thereby performing transfer learning of the information for the non-nominal instances of the one or more specimens to the machine learning based model, wherein said acquiring and said re-training are performed by the one or more computer systems, wherein one or more components are executed by the one or more computer systems, and wherein the one or more components comprise the machine learning based model.
37. A computer-implemented method for training a machine learning based model, comprising:
acquiring information for non-nominal instances of one or more specimens on which a process is performed, wherein a machine learning based model is configured for performing one or more simulations for the specimens, and wherein the machine learning based model is initially trained with only information for nominal instances of one or more additional specimens; and
re-training the machine learning based model with the information for the non-nominal instances of the one or more specimens thereby performing transfer learning of the information for the non-nominal instances of the one or more specimens to the machine learning based model, wherein said acquiring and said re-training are performed by one or more computer systems, wherein one or more components are executed by the one or more computer systems, and wherein the one or more components comprise the machine learning based model.