Patent
US 12,585,918 B2ML MODEL DRIFT DETECTION USING MODIFIED GAN
Nicole M. Hatten
Raytheon Company, Arlington, VA (US)·Mar. 24, 2026·US
Copyright © 2026 Matter42. All rights reserved.
Patent
Atlas literature
Patent
US 12,585,918 B2Patent drawings and their descriptions. Click a drawing to enlarge it.
FIG. 1 illustrates, by way of example, a diagram of a GAN architecture.
FIG. 2 illustrates, by way of example, a diagram of an embodiment of a modified GAN architecture for quantifying 40 ML model drift analytics.
FIG. 3 illustrates, by way of example, a block diagram of an ML system for model drift detection.
FIG. 4 illustrates, by way of example, a block diagram of an embodiment of a method for ML model drift detection. 45
FIG. 5 is a block diagram of an example of an environ- ment including a system for neural network training, accord- ing to an embodiment.
FIG. 6 illustrates, by way of example, a block diagram of an embodiment of a machine in the example form of a 50 computer system within which instructions, for …
Claims define the patent's legal scope. Independent claims stand alone; dependent claims (nested) narrow them. Click a claim to expand its dependents.
A device comprising: at least one memory including instructions stored thereon; and processing circuitry configured to execute the instruc-tions, the instructions, when executed, cause the pro-cessing circuitry to perform operations comprising: receiving machine learning (ML) data defining a number of layers of neurons, a number of neurons per each layer, and weights for each neuron of a deployed ML model; operating the deployed ML model in a modified genera-tive adversarial network (GAN) architecture, the modi-fied GAN architecture includes fake data, from a gen-erator of an unmodified GAN architecture that includes a discriminator, input to the deployed ML model, the modified GAN architecture includes no feedback loops configured to update weights of neurons of the modi-fied GAN architecture; while operating the deployed ML model, recording output of a hidden layer of the deployed ML model; determining a metric of the output; and re-deploying the ML model and monitoring whether the re-deployed ML model is suffering from ML model drift based on the metric.
The device of claim 1, wherein the operations further comprise training the deployed ML model based on input used to determine the metric, resulting in an ML model that does not suffer from the ML model drift.
The device of claim 1, wherein the metric is determined per class that is classified by the deployed ML model.
The device of claim 1, wherein the modified GAN architecture includes the deployed ML model configured to classify the fake data independent of real data.
A non-transitory machine-readable medium including instructions that, when executed by a machine, cause the machine to perform operations comprising: receiving machine learning (ML) data defining a number of layers of neurons, a number of neurons per each layer, and weights for each neuron of a deployed ML model; operating the deployed ML model in a modified genera-tive adversarial network (GAN) architecture, the modi-fied GAN architecture includes fake data, from a gen-erator of an unmodified GAN architecture that includes a discriminator, input to the deployed ML model, the modified GAN architecture includes no feedback loops configured to update weights of neurons of the modi-fied GAN architecture; while operating the deployed ML model, recording output of a hidden layer of the deployed ML model; determining a metric of the output; and re-deploying the deployed ML model and monitoring whether the re-deployed ML model is suffering from ML model drift based on the metric.
The non-transitory machine-readable medium of claim 6, wherein the operations further comprise training the deployed ML model based on input used to determine the metric, resulting in an ML model that does not suffer from the ML model drift.
The non-transitory machine-readable medium of claim 6, wherein the metric is determined per class that is classi-fied by the deployed ML model.
The non-transitory machine-readable medium of claim 6, wherein the modified GAN architecture includes the deployed ML model configured to classify the fake data independent of real data.
A method comprising: receiving machine learning (ML) data defining a number of layers of neurons, a number of neurons per each layer, and weights for each neuron of a deployed ML model; operating the deployed ML model in a modified genera-tive adversarial network (GAN) architecture, the modi-fied GAN architecture includes fake data, from a gen-erator of an unmodified GAN architecture that includes a discriminator, input to the deployed ML model, the modified GAN architecture includes no feedback loops configured to update weights of neurons of the modi-fied GAN architecture; while operating the deployed ML model, recording output of a hidden layer of the deployed ML model; determining a metric of the output; and re-deploying the deployed ML model and monitoring whether the re-deployed ML model is suffering from ML model drift based on the metric.
The method of claim 11, further comprising training the deployed ML model based on input used to determine the metric, resulting in an ML model that does not suffer from the ML model drift.
The method of claim 11, wherein the metric is determined per class that is classified by the deployed ML model.
The method of claim 11, wherein the modified GAN architecture includes the deployed ML model configured to classify the fake data independent of real data. ∗ ∗ ∗ ∗ ∗
Layer stacks claimed or described, ordered top of device to substrate.
ML model drift detection device
No layer stack recorded.
modified GAN architecture for ML model drift detection
No layer stack recorded.
discriminator network (deep neural network) with input layer, hidden layers, and output layer
Patents and literature cited by this patent (applicant and examiner references).
Cited patents · 2
Cited non-patent literature · 5
Patent
Atlas literature
Patent
US 12,585,918 B2Patent drawings and their descriptions. Click a drawing to enlarge it.
FIG. 1 illustrates, by way of example, a diagram of a GAN architecture.
FIG. 2 illustrates, by way of example, a diagram of an embodiment of a modified GAN architecture for quantifying 40 ML model drift analytics.
FIG. 3 illustrates, by way of example, a block diagram of an ML system for model drift detection.
FIG. 4 illustrates, by way of example, a block diagram of an embodiment of a method for ML model drift detection. 45
FIG. 5 is a block diagram of an example of an environ- ment including a system for neural network training, accord- ing to an embodiment.
FIG. 6 illustrates, by way of example, a block diagram of an embodiment of a machine in the example form of a 50 computer system within which instructions, for …
Claims define the patent's legal scope. Independent claims stand alone; dependent claims (nested) narrow them. Click a claim to expand its dependents.
A device comprising: at least one memory including instructions stored thereon; and processing circuitry configured to execute the instruc-tions, the instructions, when executed, cause the pro-cessing circuitry to perform operations comprising: receiving machine learning (ML) data defining a number of layers of neurons, a number of neurons per each layer, and weights for each neuron of a deployed ML model; operating the deployed ML model in a modified genera-tive adversarial network (GAN) architecture, the modi-fied GAN architecture includes fake data, from a gen-erator of an unmodified GAN architecture that includes a discriminator, input to the deployed ML model, the modified GAN architecture includes no feedback loops configured to update weights of neurons of the modi-fied GAN architecture; while operating the deployed ML model, recording output of a hidden layer of the deployed ML model; determining a metric of the output; and re-deploying the ML model and monitoring whether the re-deployed ML model is suffering from ML model drift based on the metric.
The device of claim 1, wherein the operations further comprise training the deployed ML model based on input used to determine the metric, resulting in an ML model that does not suffer from the ML model drift.
The device of claim 1, wherein the metric is determined per class that is classified by the deployed ML model.
The device of claim 1, wherein the modified GAN architecture includes the deployed ML model configured to classify the fake data independent of real data.
A non-transitory machine-readable medium including instructions that, when executed by a machine, cause the machine to perform operations comprising: receiving machine learning (ML) data defining a number of layers of neurons, a number of neurons per each layer, and weights for each neuron of a deployed ML model; operating the deployed ML model in a modified genera-tive adversarial network (GAN) architecture, the modi-fied GAN architecture includes fake data, from a gen-erator of an unmodified GAN architecture that includes a discriminator, input to the deployed ML model, the modified GAN architecture includes no feedback loops configured to update weights of neurons of the modi-fied GAN architecture; while operating the deployed ML model, recording output of a hidden layer of the deployed ML model; determining a metric of the output; and re-deploying the deployed ML model and monitoring whether the re-deployed ML model is suffering from ML model drift based on the metric.
The non-transitory machine-readable medium of claim 6, wherein the operations further comprise training the deployed ML model based on input used to determine the metric, resulting in an ML model that does not suffer from the ML model drift.
The non-transitory machine-readable medium of claim 6, wherein the metric is determined per class that is classi-fied by the deployed ML model.
The non-transitory machine-readable medium of claim 6, wherein the modified GAN architecture includes the deployed ML model configured to classify the fake data independent of real data.
A method comprising: receiving machine learning (ML) data defining a number of layers of neurons, a number of neurons per each layer, and weights for each neuron of a deployed ML model; operating the deployed ML model in a modified genera-tive adversarial network (GAN) architecture, the modi-fied GAN architecture includes fake data, from a gen-erator of an unmodified GAN architecture that includes a discriminator, input to the deployed ML model, the modified GAN architecture includes no feedback loops configured to update weights of neurons of the modi-fied GAN architecture; while operating the deployed ML model, recording output of a hidden layer of the deployed ML model; determining a metric of the output; and re-deploying the deployed ML model and monitoring whether the re-deployed ML model is suffering from ML model drift based on the metric.
The method of claim 11, further comprising training the deployed ML model based on input used to determine the metric, resulting in an ML model that does not suffer from the ML model drift.
The method of claim 11, wherein the metric is determined per class that is classified by the deployed ML model.
The method of claim 11, wherein the modified GAN architecture includes the deployed ML model configured to classify the fake data independent of real data. ∗ ∗ ∗ ∗ ∗
Layer stacks claimed or described, ordered top of device to substrate.
ML model drift detection device
No layer stack recorded.
modified GAN architecture for ML model drift detection
No layer stack recorded.
discriminator network (deep neural network) with input layer, hidden layers, and output layer
Patents and literature cited by this patent (applicant and examiner references).
Cited patents · 2
Cited non-patent literature · 5
Patent
Atlas literature
Patent
US 12,585,918 B2Patent drawings and their descriptions. Click a drawing to enlarge it.
FIG. 1 illustrates, by way of example, a diagram of a GAN architecture.
FIG. 2 illustrates, by way of example, a diagram of an embodiment of a modified GAN architecture for quantifying 40 ML model drift analytics.
FIG. 3 illustrates, by way of example, a block diagram of an ML system for model drift detection.
FIG. 4 illustrates, by way of example, a block diagram of an embodiment of a method for ML model drift detection. 45
FIG. 5 is a block diagram of an example of an environ- ment including a system for neural network training, accord- ing to an embodiment.
FIG. 6 illustrates, by way of example, a block diagram of an embodiment of a machine in the example form of a 50 computer system within which instructions, for …
Claims define the patent's legal scope. Independent claims stand alone; dependent claims (nested) narrow them. Click a claim to expand its dependents.
A device comprising: at least one memory including instructions stored thereon; and processing circuitry configured to execute the instruc-tions, the instructions, when executed, cause the pro-cessing circuitry to perform operations comprising: receiving machine learning (ML) data defining a number of layers of neurons, a number of neurons per each layer, and weights for each neuron of a deployed ML model; operating the deployed ML model in a modified genera-tive adversarial network (GAN) architecture, the modi-fied GAN architecture includes fake data, from a gen-erator of an unmodified GAN architecture that includes a discriminator, input to the deployed ML model, the modified GAN architecture includes no feedback loops configured to update weights of neurons of the modi-fied GAN architecture; while operating the deployed ML model, recording output of a hidden layer of the deployed ML model; determining a metric of the output; and re-deploying the ML model and monitoring whether the re-deployed ML model is suffering from ML model drift based on the metric.
The device of claim 1, wherein the operations further comprise training the deployed ML model based on input used to determine the metric, resulting in an ML model that does not suffer from the ML model drift.
The device of claim 1, wherein the metric is determined per class that is classified by the deployed ML model.
The device of claim 1, wherein the modified GAN architecture includes the deployed ML model configured to classify the fake data independent of real data.
A non-transitory machine-readable medium including instructions that, when executed by a machine, cause the machine to perform operations comprising: receiving machine learning (ML) data defining a number of layers of neurons, a number of neurons per each layer, and weights for each neuron of a deployed ML model; operating the deployed ML model in a modified genera-tive adversarial network (GAN) architecture, the modi-fied GAN architecture includes fake data, from a gen-erator of an unmodified GAN architecture that includes a discriminator, input to the deployed ML model, the modified GAN architecture includes no feedback loops configured to update weights of neurons of the modi-fied GAN architecture; while operating the deployed ML model, recording output of a hidden layer of the deployed ML model; determining a metric of the output; and re-deploying the deployed ML model and monitoring whether the re-deployed ML model is suffering from ML model drift based on the metric.
The non-transitory machine-readable medium of claim 6, wherein the operations further comprise training the deployed ML model based on input used to determine the metric, resulting in an ML model that does not suffer from the ML model drift.
The non-transitory machine-readable medium of claim 6, wherein the metric is determined per class that is classi-fied by the deployed ML model.
The non-transitory machine-readable medium of claim 6, wherein the modified GAN architecture includes the deployed ML model configured to classify the fake data independent of real data.
A method comprising: receiving machine learning (ML) data defining a number of layers of neurons, a number of neurons per each layer, and weights for each neuron of a deployed ML model; operating the deployed ML model in a modified genera-tive adversarial network (GAN) architecture, the modi-fied GAN architecture includes fake data, from a gen-erator of an unmodified GAN architecture that includes a discriminator, input to the deployed ML model, the modified GAN architecture includes no feedback loops configured to update weights of neurons of the modi-fied GAN architecture; while operating the deployed ML model, recording output of a hidden layer of the deployed ML model; determining a metric of the output; and re-deploying the deployed ML model and monitoring whether the re-deployed ML model is suffering from ML model drift based on the metric.
The method of claim 11, further comprising training the deployed ML model based on input used to determine the metric, resulting in an ML model that does not suffer from the ML model drift.
The method of claim 11, wherein the metric is determined per class that is classified by the deployed ML model.
The method of claim 11, wherein the modified GAN architecture includes the deployed ML model configured to classify the fake data independent of real data. ∗ ∗ ∗ ∗ ∗
Layer stacks claimed or described, ordered top of device to substrate.
ML model drift detection device
No layer stack recorded.
modified GAN architecture for ML model drift detection
No layer stack recorded.
discriminator network (deep neural network) with input layer, hidden layers, and output layer
Patents and literature cited by this patent (applicant and examiner references).
Cited patents · 2
Cited non-patent literature · 5
Patent
Atlas literature
Patent
US 12,585,918 B2Patent drawings and their descriptions. Click a drawing to enlarge it.
FIG. 1 illustrates, by way of example, a diagram of a GAN architecture.
FIG. 2 illustrates, by way of example, a diagram of an embodiment of a modified GAN architecture for quantifying 40 ML model drift analytics.
FIG. 3 illustrates, by way of example, a block diagram of an ML system for model drift detection.
FIG. 4 illustrates, by way of example, a block diagram of an embodiment of a method for ML model drift detection. 45
FIG. 5 is a block diagram of an example of an environ- ment including a system for neural network training, accord- ing to an embodiment.
FIG. 6 illustrates, by way of example, a block diagram of an embodiment of a machine in the example form of a 50 computer system within which instructions, for …
Claims define the patent's legal scope. Independent claims stand alone; dependent claims (nested) narrow them. Click a claim to expand its dependents.
A device comprising: at least one memory including instructions stored thereon; and processing circuitry configured to execute the instruc-tions, the instructions, when executed, cause the pro-cessing circuitry to perform operations comprising: receiving machine learning (ML) data defining a number of layers of neurons, a number of neurons per each layer, and weights for each neuron of a deployed ML model; operating the deployed ML model in a modified genera-tive adversarial network (GAN) architecture, the modi-fied GAN architecture includes fake data, from a gen-erator of an unmodified GAN architecture that includes a discriminator, input to the deployed ML model, the modified GAN architecture includes no feedback loops configured to update weights of neurons of the modi-fied GAN architecture; while operating the deployed ML model, recording output of a hidden layer of the deployed ML model; determining a metric of the output; and re-deploying the ML model and monitoring whether the re-deployed ML model is suffering from ML model drift based on the metric.
The device of claim 1, wherein the operations further comprise training the deployed ML model based on input used to determine the metric, resulting in an ML model that does not suffer from the ML model drift.
The device of claim 1, wherein the metric is determined per class that is classified by the deployed ML model.
The device of claim 1, wherein the modified GAN architecture includes the deployed ML model configured to classify the fake data independent of real data.
A non-transitory machine-readable medium including instructions that, when executed by a machine, cause the machine to perform operations comprising: receiving machine learning (ML) data defining a number of layers of neurons, a number of neurons per each layer, and weights for each neuron of a deployed ML model; operating the deployed ML model in a modified genera-tive adversarial network (GAN) architecture, the modi-fied GAN architecture includes fake data, from a gen-erator of an unmodified GAN architecture that includes a discriminator, input to the deployed ML model, the modified GAN architecture includes no feedback loops configured to update weights of neurons of the modi-fied GAN architecture; while operating the deployed ML model, recording output of a hidden layer of the deployed ML model; determining a metric of the output; and re-deploying the deployed ML model and monitoring whether the re-deployed ML model is suffering from ML model drift based on the metric.
The non-transitory machine-readable medium of claim 6, wherein the operations further comprise training the deployed ML model based on input used to determine the metric, resulting in an ML model that does not suffer from the ML model drift.
The non-transitory machine-readable medium of claim 6, wherein the metric is determined per class that is classi-fied by the deployed ML model.
The non-transitory machine-readable medium of claim 6, wherein the modified GAN architecture includes the deployed ML model configured to classify the fake data independent of real data.
A method comprising: receiving machine learning (ML) data defining a number of layers of neurons, a number of neurons per each layer, and weights for each neuron of a deployed ML model; operating the deployed ML model in a modified genera-tive adversarial network (GAN) architecture, the modi-fied GAN architecture includes fake data, from a gen-erator of an unmodified GAN architecture that includes a discriminator, input to the deployed ML model, the modified GAN architecture includes no feedback loops configured to update weights of neurons of the modi-fied GAN architecture; while operating the deployed ML model, recording output of a hidden layer of the deployed ML model; determining a metric of the output; and re-deploying the deployed ML model and monitoring whether the re-deployed ML model is suffering from ML model drift based on the metric.
The method of claim 11, further comprising training the deployed ML model based on input used to determine the metric, resulting in an ML model that does not suffer from the ML model drift.
The method of claim 11, wherein the metric is determined per class that is classified by the deployed ML model.
The method of claim 11, wherein the modified GAN architecture includes the deployed ML model configured to classify the fake data independent of real data. ∗ ∗ ∗ ∗ ∗
Layer stacks claimed or described, ordered top of device to substrate.
ML model drift detection device
No layer stack recorded.
modified GAN architecture for ML model drift detection
No layer stack recorded.
discriminator network (deep neural network) with input layer, hidden layers, and output layer
Patents and literature cited by this patent (applicant and examiner references).
Cited patents · 2
Cited non-patent literature · 5