Machine learning techniques for wafer defect map classification
Various embodiments of the present disclosure provide methods, apparatus, systems, computing devices, computing entities for generating classification predictions for wafer defect maps. Reduced feature data may be generated, using a non-linear dimensionality reduction machine learning model and based at least in part on vector representations for a set of wafer defect map images. One or more wafer defect pattern clusters may be generated, using a density-based clustering machine learning model and based at least in part on the reduced feature data. Each wafer defect map image may be associated with a particular wafer defect pattern cluster of the one or more wafer defect pattern clusters. A classification prediction may be generated for each wafer defect map image based at least in part on the particular wafer defect pattern cluster associated with the respective wafer defect map image.
1 . A device, comprising:
one or more processors; and
an unsupervised classification circuitry, configured to execute one or more instructions to:
generate, using a non-linear dimensionality reduction machine learning model and based at least in part on vector representations for a set of wafer defect map images, reduced feature data, wherein the reduced feature data comprises feature data for the set of wafer defect map images in a reduced dimensionality space;
generate, using a density-based clustering machine learning model and based at least in part on the reduced feature data, one or more wafer defect pattern clusters, wherein each wafer defect map image is associated with a particular wafer defect pattern cluster of the one or more wafer defect pattern clusters;
generate a classification prediction for each wafer defect map image based at least in part on the particular wafer defect pattern cluster associated with the respective wafer defect map image;
identify a cluster representative for each wafer defect pattern cluster of the one or more wafer defect pattern clusters; and
for each cluster representative:
generate, based at least in part on a set of candidate class representatives, one or more pattern similarity scores, wherein:
each pattern similarity score corresponds to a representative pair of one or more representative pairs, and
each representative pair comprises the cluster representative and a particular candidate class representative from the set of candidate class representatives; and
update based at least in part on association of each pattern similarity score with a non-qualifying pattern similarity score, a wafer defect pattern class corpus.
2 . The device of claim 1 , wherein the set of wafer defect map images comprise preprocessed wafer defect map images.
3 . The device of claim 2 , wherein the unsupervised classification circuitry, is configured to execute the one or more instructions to generate the preprocessed wafer defect map images by (i) reducing a spatial dimension for each wafer defect map image in the set of wafer defect map images, and (ii) performing one or more filtering operations to remove non-qualifying wafer defect map images from the set of wafer defect map images.
4 . The device of claim 1 , wherein the unsupervised classification circuitry is configured to execute the one or more instructions to generate the vector representations for the set of wafer defect map images based at least in part by performing a flattening operation on each wafer defect map image in the set of wafer defect map images.
5 . The device of claim 1 , wherein:
the non-linear dimensionality reduction machine learning model is a Uniform Manifold Approximation and Projection machine learning model; and
the unsupervised classification circuitry is configured to execute the one or more instructions to generate the reduced feature data by performing Uniform Manifold Approximation and Projection on the vector representations for the set of wafer defect map images.
6 . The device of claim 1 , wherein:
the density-based clustering machine learning model is a Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) machine learning model.
7 . The device of claim 1 , wherein the unsupervised classification circuitry is configured to execute the one or more instructions to generate the classification prediction for each wafer defect map image by (i) assigning a label to each wafer defect pattern cluster, and (ii) associating the respective wafer defect map image with the label for the wafer defect pattern cluster associated with the respective wafer defect map.
8 . The device of claim 1 , wherein the unsupervised classification circuitry is configured to execute the one or more instructions to update the wafer defect pattern class corpus by updating a wafer defect pattern class database.
9 . The device of claim 1 , further comprising:
a supervised classification circuitry, configured to execute one or more instructions to:
train, based at least in part on the updated wafer defect pattern class corpus, a convolutional neural network machine learning model configured to generate classification predictions for wafer defect maps.
10 . A system, comprising:
one or more memories; and
an unsupervised classification circuitry, configured to execute one or more instructions to:
generate, using a non-linear dimensionality reduction machine learning model and based at least in part on vector representations for a set of wafer defect map images, reduced feature data, wherein the reduced feature data comprises feature data for the set of wafer defect map images in a reduced dimensionality space;
generate, using a density-based clustering machine learning model and based at least in part on the reduced feature data, one or more wafer defect pattern clusters, wherein each wafer defect map image is associated with a particular wafer defect pattern cluster of the one or more wafer defect pattern clusters;
generate a classification prediction for each wafer defect map image based at least in part on the particular wafer defect pattern cluster associated with the respective wafer defect map image;
identify a cluster representative for each wafer defect pattern cluster of the one or more wafer defect pattern clusters; and
for each cluster representative:
generate, based at least in part on a set of candidate class representatives, one or more pattern similarity scores, wherein:
each pattern similarity score corresponds to a representative pair of one or more representative pairs, and
each representative pair comprises the cluster representative and a particular candidate class representative from the set of candidate class representatives; and
update based at least in part on association of each pattern similarity score with a non-qualifying pattern similarity score, a wafer defect pattern class corpus.
11 . The system of claim 10 , wherein the set of wafer defect map images comprises preprocessed wafer defect map images.
12 . The system of claim 11 , wherein the unsupervised classification circuitry, is configured to execute the one or more instructions to generate the preprocessed wafer defect map images by: (i) reducing a spatial dimension for each wafer defect map image in the set of wafer defect map images, and (ii) performing one or more filtering operations to remove non-qualifying wafer defect map images from the set of wafer defect map images.
13 . The system of claim 10 , wherein the unsupervised classification circuitry is configured to execute one or more instructions to generate the vector representations for the set of wafer defect map images based at least in part by performing a flattening operation on each wafer defect map image in the set of wafer defect map images.
14 . The system of claim 10 , wherein:
the non-linear dimensionality reduction machine learning model is a Uniform Manifold Approximation and Projection machine learning model; and
the unsupervised classification circuitry is configured to execute the one or more instructions to generate the reduced feature data by performing Uniform Manifold Approximation and Projection on the vector representations for the set of wafer defect map images.
15 . The system of claim 10 , wherein:
the density-based clustering machine learning model is a Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) machine learning model.
16 . The system of claim 10 , wherein the unsupervised classification circuitry is configured to execute the one or more instructions to generate the classification prediction for each wafer defect map image by (i) assigning a label to each wafer defect pattern cluster, and (ii) associating the respective wafer defect map image with the label for the wafer defect pattern cluster associated with the respective wafer defect map.
17 . The system of claim 10 , wherein the unsupervised classification circuitry is configured to execute the one or more instructions to update the wafer defect pattern class corpus by updating a wafer defect pattern class database.
18 . The system of claim 10 , further comprising:
a supervised classification circuitry, configured to execute one or more instructions to:
train, based at least in part on the updated wafer defect pattern class corpus, a convolutional neural network machine learning model configured to generate classification predictions for wafer defect maps.