Patent
Atlas literature
Patent
US 12,248,556 B2Patent drawings and their descriptions. Click a drawing to enlarge it.
FIG. 1 is a block diagram of an example computing system that may be used to provide an authenticator-inte- gration generative adversarial network (GAN) for …
FIG. 2 illustrates a machine learning software stack, according to an embodiment.
FIGS. 3A-3B illustrate layers of example deep neural 60 networks.
FIG. 4 illustrates an example recurrent neural network.
FIG. 5 illustrates training and deployment of a deep neural network.
FIG. 6 depicts a GAN system providing an authenticator- 65 integrated GAN for secure deepfake generation, in accor- dance with implementations of the …
FIG. 7 depicts a schematic of an illustrative flow of an authenticator-integrated GAN system providing secure deepfake detection, in accordance with …
FIG. 8 is a flow diagram illustrating an embodiment of a method for a secure deepfake generation using an authen- ticator-integrated GAN.
FIG. 9 is a flow diagram illustrating an embodiment of a method 900 for training an authenticator-integrated GAN for secure deepfake generation.
FIG. 10 is a schematic diagram of an illustrative elec- tronic computing device to enable an authenticator-integra- tion GAN for secure deepfake generation, …
Claims define the patent's legal scope. Independent claims stand alone; dependent claims (nested) narrow them. Click a claim to expand its dependents.
An apparatus comprising: one or more processors to: generate, by a generative neural network, samples based on feedback received from a discriminator neural network and from an authenticator neural network, the generative neural network is to trick the discriminator neural network to identify the gener-ated samples as real content samples; digest, by the authenticator neural network, the real content samples, the generated samples from the generative neural network, and an authentication code; embed, by the authenticator neural network, the authen-tication code into the generated samples from the generative neural network as an embedded authen-tication code by contributing to a generator loss provided to the generative neural network; generate, by the generative neural network, content comprising the embedded authentication code; and verify, by the authenticator neural network, the content based on the embedded authentication code.
The apparatus of claim 1, wherein a combination of the generative neural network, the discriminator neural network, and the authenticator neural network comprise a generative adversarial network (GAN).
The apparatus of claim 1, wherein the authentication code comprises at least one of a hash of a signal or a password to create a string of characters.
The apparatus of claim 1, wherein the authentication code comprises an encoder to create authentication codes from given strings.
The apparatus of claim 1, wherein the generative neural network to generate adversarial loss comprising a first loss function between the generative neural network and the discriminator neural network, and wherein the first loss function is to characterize whether the content from the generative neural network is following a first distribution of original content input to the generative neural network.
The apparatus of claim 1, wherein the embedded authentication code is decoded and verified as authentic using a learned representation in the authenticator neural network, and wherein to authenticate the content, the authenticator neural network to decode the content created using the authentication code.
A non-transitory computer-readable storage medium having stored thereon executable computer program instruc-tions that, when executed by one or more processors, cause the one or more processors to perform operations compris-ing: generating, by a generative neural network, samples based on feedback received from a discriminator neural net-work and from an authenticator neural network, the generative neural network is to trick the discriminator neural network to identify the generated samples as real content samples; digesting, by the authenticator neural network, the real content samples, the generated samples from the gen-erative neural network, and an authentication code; embedding, by the authenticator neural network, the authentication code into the generated samples from the generative neural network as an embedded authentica-tion code by contributing to a generator loss provided to the generative neural network; generating, by the generative neural network, content comprising the embedded authentication code; and verifying, by the authenticator neural network, the content based on the embedded authentication code; wherein a combination of the generative neural network, the discriminator neural network, and the authenticator neural network comprise a generative adversarial net-work (GAN).
The non-transitory computer-readable storage medium of claim 10, wherein the generative neural network to generate adversarial loss comprising a first loss function between the generative neural network and the discriminator neural network, and wherein the first loss function is to characterize whether the content from the generative neural network is following a first distribution of original content input to the generative neural network.
The non-transitory computer-readable storage medium of claim 10, wherein the embedded authentication code is decoded and verified as authentic using a learned representation in the authenticator neural network, and wherein to authenticate the content, the authenticator neural network decodes the content created using the authentication code.
A method comprising: generating, by a processor using a generative neural network, samples based on feedback received from a discriminator neural network and from an authenticator neural network, the generative neural network is to trick the discriminator neural network to identify the generated samples as real content samples; digesting, by the processor using the authenticator neural network, the real content samples, the generated samples from the generative neural network, and an authentication code; embedding, by the processor using the authenticator neu-ral network, the authentication code into the generated samples from the generative neural network as an embedded authentication code by contributing to a generator loss provided to the generative neural net-work; generating, by the processor using the generative neural network, content comprising the embedded authentica-tion code; and verifying, by the processor using the authenticator neural network, the content based on the embedded authenti-cation code; wherein a combination of the generative neural network, the discriminator neural network, and the authenticator neural network comprise a generative adversarial net-work (GAN).
The method of claim 16, wherein the generative neural network to generate adversarial loss comprising a first loss function between the generative neural network and the discriminator neural network, and wherein the first loss function is to characterize whether the content from the generative neural network is following a first distribution of original content that is input to the generative neural net-work.
Layer stacks claimed or described, ordered top of device to substrate.
authenticator-integrated GAN apparatus
No layer stack recorded.
non-transitory computer-readable storage medium with authenticator-integrated GAN instructions
No layer stack recorded.
authenticator-integrated GAN method executed by processor
No layer stack recorded.
Patents and literature cited by this patent (applicant and examiner references).
Cited patents · 6
Patent
Atlas literature
Patent
US 12,248,556 B2Patent drawings and their descriptions. Click a drawing to enlarge it.
FIG. 1 is a block diagram of an example computing system that may be used to provide an authenticator-inte- gration generative adversarial network (GAN) for …
FIG. 2 illustrates a machine learning software stack, according to an embodiment.
FIGS. 3A-3B illustrate layers of example deep neural 60 networks.
FIG. 4 illustrates an example recurrent neural network.
FIG. 5 illustrates training and deployment of a deep neural network.
FIG. 6 depicts a GAN system providing an authenticator- 65 integrated GAN for secure deepfake generation, in accor- dance with implementations of the …
FIG. 7 depicts a schematic of an illustrative flow of an authenticator-integrated GAN system providing secure deepfake detection, in accordance with …
FIG. 8 is a flow diagram illustrating an embodiment of a method for a secure deepfake generation using an authen- ticator-integrated GAN.
FIG. 9 is a flow diagram illustrating an embodiment of a method 900 for training an authenticator-integrated GAN for secure deepfake generation.
FIG. 10 is a schematic diagram of an illustrative elec- tronic computing device to enable an authenticator-integra- tion GAN for secure deepfake generation, …
Claims define the patent's legal scope. Independent claims stand alone; dependent claims (nested) narrow them. Click a claim to expand its dependents.
An apparatus comprising: one or more processors to: generate, by a generative neural network, samples based on feedback received from a discriminator neural network and from an authenticator neural network, the generative neural network is to trick the discriminator neural network to identify the gener-ated samples as real content samples; digest, by the authenticator neural network, the real content samples, the generated samples from the generative neural network, and an authentication code; embed, by the authenticator neural network, the authen-tication code into the generated samples from the generative neural network as an embedded authen-tication code by contributing to a generator loss provided to the generative neural network; generate, by the generative neural network, content comprising the embedded authentication code; and verify, by the authenticator neural network, the content based on the embedded authentication code.
The apparatus of claim 1, wherein a combination of the generative neural network, the discriminator neural network, and the authenticator neural network comprise a generative adversarial network (GAN).
The apparatus of claim 1, wherein the authentication code comprises at least one of a hash of a signal or a password to create a string of characters.
The apparatus of claim 1, wherein the authentication code comprises an encoder to create authentication codes from given strings.
The apparatus of claim 1, wherein the generative neural network to generate adversarial loss comprising a first loss function between the generative neural network and the discriminator neural network, and wherein the first loss function is to characterize whether the content from the generative neural network is following a first distribution of original content input to the generative neural network.
The apparatus of claim 1, wherein the embedded authentication code is decoded and verified as authentic using a learned representation in the authenticator neural network, and wherein to authenticate the content, the authenticator neural network to decode the content created using the authentication code.
A non-transitory computer-readable storage medium having stored thereon executable computer program instruc-tions that, when executed by one or more processors, cause the one or more processors to perform operations compris-ing: generating, by a generative neural network, samples based on feedback received from a discriminator neural net-work and from an authenticator neural network, the generative neural network is to trick the discriminator neural network to identify the generated samples as real content samples; digesting, by the authenticator neural network, the real content samples, the generated samples from the gen-erative neural network, and an authentication code; embedding, by the authenticator neural network, the authentication code into the generated samples from the generative neural network as an embedded authentica-tion code by contributing to a generator loss provided to the generative neural network; generating, by the generative neural network, content comprising the embedded authentication code; and verifying, by the authenticator neural network, the content based on the embedded authentication code; wherein a combination of the generative neural network, the discriminator neural network, and the authenticator neural network comprise a generative adversarial net-work (GAN).
The non-transitory computer-readable storage medium of claim 10, wherein the generative neural network to generate adversarial loss comprising a first loss function between the generative neural network and the discriminator neural network, and wherein the first loss function is to characterize whether the content from the generative neural network is following a first distribution of original content input to the generative neural network.
The non-transitory computer-readable storage medium of claim 10, wherein the embedded authentication code is decoded and verified as authentic using a learned representation in the authenticator neural network, and wherein to authenticate the content, the authenticator neural network decodes the content created using the authentication code.
A method comprising: generating, by a processor using a generative neural network, samples based on feedback received from a discriminator neural network and from an authenticator neural network, the generative neural network is to trick the discriminator neural network to identify the generated samples as real content samples; digesting, by the processor using the authenticator neural network, the real content samples, the generated samples from the generative neural network, and an authentication code; embedding, by the processor using the authenticator neu-ral network, the authentication code into the generated samples from the generative neural network as an embedded authentication code by contributing to a generator loss provided to the generative neural net-work; generating, by the processor using the generative neural network, content comprising the embedded authentica-tion code; and verifying, by the processor using the authenticator neural network, the content based on the embedded authenti-cation code; wherein a combination of the generative neural network, the discriminator neural network, and the authenticator neural network comprise a generative adversarial net-work (GAN).
The method of claim 16, wherein the generative neural network to generate adversarial loss comprising a first loss function between the generative neural network and the discriminator neural network, and wherein the first loss function is to characterize whether the content from the generative neural network is following a first distribution of original content that is input to the generative neural net-work.
Layer stacks claimed or described, ordered top of device to substrate.
authenticator-integrated GAN apparatus
No layer stack recorded.
non-transitory computer-readable storage medium with authenticator-integrated GAN instructions
No layer stack recorded.
authenticator-integrated GAN method executed by processor
No layer stack recorded.
Patents and literature cited by this patent (applicant and examiner references).
Cited patents · 6
Patent
Atlas literature
Patent
US 12,248,556 B2Patent drawings and their descriptions. Click a drawing to enlarge it.
FIG. 1 is a block diagram of an example computing system that may be used to provide an authenticator-inte- gration generative adversarial network (GAN) for …
FIG. 2 illustrates a machine learning software stack, according to an embodiment.
FIGS. 3A-3B illustrate layers of example deep neural 60 networks.
FIG. 4 illustrates an example recurrent neural network.
FIG. 5 illustrates training and deployment of a deep neural network.
FIG. 6 depicts a GAN system providing an authenticator- 65 integrated GAN for secure deepfake generation, in accor- dance with implementations of the …
FIG. 7 depicts a schematic of an illustrative flow of an authenticator-integrated GAN system providing secure deepfake detection, in accordance with …
FIG. 8 is a flow diagram illustrating an embodiment of a method for a secure deepfake generation using an authen- ticator-integrated GAN.
FIG. 9 is a flow diagram illustrating an embodiment of a method 900 for training an authenticator-integrated GAN for secure deepfake generation.
FIG. 10 is a schematic diagram of an illustrative elec- tronic computing device to enable an authenticator-integra- tion GAN for secure deepfake generation, …
Claims define the patent's legal scope. Independent claims stand alone; dependent claims (nested) narrow them. Click a claim to expand its dependents.
An apparatus comprising: one or more processors to: generate, by a generative neural network, samples based on feedback received from a discriminator neural network and from an authenticator neural network, the generative neural network is to trick the discriminator neural network to identify the gener-ated samples as real content samples; digest, by the authenticator neural network, the real content samples, the generated samples from the generative neural network, and an authentication code; embed, by the authenticator neural network, the authen-tication code into the generated samples from the generative neural network as an embedded authen-tication code by contributing to a generator loss provided to the generative neural network; generate, by the generative neural network, content comprising the embedded authentication code; and verify, by the authenticator neural network, the content based on the embedded authentication code.
The apparatus of claim 1, wherein a combination of the generative neural network, the discriminator neural network, and the authenticator neural network comprise a generative adversarial network (GAN).
The apparatus of claim 1, wherein the authentication code comprises at least one of a hash of a signal or a password to create a string of characters.
The apparatus of claim 1, wherein the authentication code comprises an encoder to create authentication codes from given strings.
The apparatus of claim 1, wherein the generative neural network to generate adversarial loss comprising a first loss function between the generative neural network and the discriminator neural network, and wherein the first loss function is to characterize whether the content from the generative neural network is following a first distribution of original content input to the generative neural network.
The apparatus of claim 1, wherein the embedded authentication code is decoded and verified as authentic using a learned representation in the authenticator neural network, and wherein to authenticate the content, the authenticator neural network to decode the content created using the authentication code.
A non-transitory computer-readable storage medium having stored thereon executable computer program instruc-tions that, when executed by one or more processors, cause the one or more processors to perform operations compris-ing: generating, by a generative neural network, samples based on feedback received from a discriminator neural net-work and from an authenticator neural network, the generative neural network is to trick the discriminator neural network to identify the generated samples as real content samples; digesting, by the authenticator neural network, the real content samples, the generated samples from the gen-erative neural network, and an authentication code; embedding, by the authenticator neural network, the authentication code into the generated samples from the generative neural network as an embedded authentica-tion code by contributing to a generator loss provided to the generative neural network; generating, by the generative neural network, content comprising the embedded authentication code; and verifying, by the authenticator neural network, the content based on the embedded authentication code; wherein a combination of the generative neural network, the discriminator neural network, and the authenticator neural network comprise a generative adversarial net-work (GAN).
The non-transitory computer-readable storage medium of claim 10, wherein the generative neural network to generate adversarial loss comprising a first loss function between the generative neural network and the discriminator neural network, and wherein the first loss function is to characterize whether the content from the generative neural network is following a first distribution of original content input to the generative neural network.
The non-transitory computer-readable storage medium of claim 10, wherein the embedded authentication code is decoded and verified as authentic using a learned representation in the authenticator neural network, and wherein to authenticate the content, the authenticator neural network decodes the content created using the authentication code.
A method comprising: generating, by a processor using a generative neural network, samples based on feedback received from a discriminator neural network and from an authenticator neural network, the generative neural network is to trick the discriminator neural network to identify the generated samples as real content samples; digesting, by the processor using the authenticator neural network, the real content samples, the generated samples from the generative neural network, and an authentication code; embedding, by the processor using the authenticator neu-ral network, the authentication code into the generated samples from the generative neural network as an embedded authentication code by contributing to a generator loss provided to the generative neural net-work; generating, by the processor using the generative neural network, content comprising the embedded authentica-tion code; and verifying, by the processor using the authenticator neural network, the content based on the embedded authenti-cation code; wherein a combination of the generative neural network, the discriminator neural network, and the authenticator neural network comprise a generative adversarial net-work (GAN).
The method of claim 16, wherein the generative neural network to generate adversarial loss comprising a first loss function between the generative neural network and the discriminator neural network, and wherein the first loss function is to characterize whether the content from the generative neural network is following a first distribution of original content that is input to the generative neural net-work.
Layer stacks claimed or described, ordered top of device to substrate.
authenticator-integrated GAN apparatus
No layer stack recorded.
non-transitory computer-readable storage medium with authenticator-integrated GAN instructions
No layer stack recorded.
authenticator-integrated GAN method executed by processor
No layer stack recorded.
Patents and literature cited by this patent (applicant and examiner references).
Cited patents · 6
Patent
Atlas literature
Patent
US 12,248,556 B2Patent drawings and their descriptions. Click a drawing to enlarge it.
FIG. 1 is a block diagram of an example computing system that may be used to provide an authenticator-inte- gration generative adversarial network (GAN) for …
FIG. 2 illustrates a machine learning software stack, according to an embodiment.
FIGS. 3A-3B illustrate layers of example deep neural 60 networks.
FIG. 4 illustrates an example recurrent neural network.
FIG. 5 illustrates training and deployment of a deep neural network.
FIG. 6 depicts a GAN system providing an authenticator- 65 integrated GAN for secure deepfake generation, in accor- dance with implementations of the …
FIG. 7 depicts a schematic of an illustrative flow of an authenticator-integrated GAN system providing secure deepfake detection, in accordance with …
FIG. 8 is a flow diagram illustrating an embodiment of a method for a secure deepfake generation using an authen- ticator-integrated GAN.
FIG. 9 is a flow diagram illustrating an embodiment of a method 900 for training an authenticator-integrated GAN for secure deepfake generation.
FIG. 10 is a schematic diagram of an illustrative elec- tronic computing device to enable an authenticator-integra- tion GAN for secure deepfake generation, …
Claims define the patent's legal scope. Independent claims stand alone; dependent claims (nested) narrow them. Click a claim to expand its dependents.
An apparatus comprising: one or more processors to: generate, by a generative neural network, samples based on feedback received from a discriminator neural network and from an authenticator neural network, the generative neural network is to trick the discriminator neural network to identify the gener-ated samples as real content samples; digest, by the authenticator neural network, the real content samples, the generated samples from the generative neural network, and an authentication code; embed, by the authenticator neural network, the authen-tication code into the generated samples from the generative neural network as an embedded authen-tication code by contributing to a generator loss provided to the generative neural network; generate, by the generative neural network, content comprising the embedded authentication code; and verify, by the authenticator neural network, the content based on the embedded authentication code.
The apparatus of claim 1, wherein a combination of the generative neural network, the discriminator neural network, and the authenticator neural network comprise a generative adversarial network (GAN).
The apparatus of claim 1, wherein the authentication code comprises at least one of a hash of a signal or a password to create a string of characters.
The apparatus of claim 1, wherein the authentication code comprises an encoder to create authentication codes from given strings.
The apparatus of claim 1, wherein the generative neural network to generate adversarial loss comprising a first loss function between the generative neural network and the discriminator neural network, and wherein the first loss function is to characterize whether the content from the generative neural network is following a first distribution of original content input to the generative neural network.
The apparatus of claim 1, wherein the embedded authentication code is decoded and verified as authentic using a learned representation in the authenticator neural network, and wherein to authenticate the content, the authenticator neural network to decode the content created using the authentication code.
A non-transitory computer-readable storage medium having stored thereon executable computer program instruc-tions that, when executed by one or more processors, cause the one or more processors to perform operations compris-ing: generating, by a generative neural network, samples based on feedback received from a discriminator neural net-work and from an authenticator neural network, the generative neural network is to trick the discriminator neural network to identify the generated samples as real content samples; digesting, by the authenticator neural network, the real content samples, the generated samples from the gen-erative neural network, and an authentication code; embedding, by the authenticator neural network, the authentication code into the generated samples from the generative neural network as an embedded authentica-tion code by contributing to a generator loss provided to the generative neural network; generating, by the generative neural network, content comprising the embedded authentication code; and verifying, by the authenticator neural network, the content based on the embedded authentication code; wherein a combination of the generative neural network, the discriminator neural network, and the authenticator neural network comprise a generative adversarial net-work (GAN).
The non-transitory computer-readable storage medium of claim 10, wherein the generative neural network to generate adversarial loss comprising a first loss function between the generative neural network and the discriminator neural network, and wherein the first loss function is to characterize whether the content from the generative neural network is following a first distribution of original content input to the generative neural network.
The non-transitory computer-readable storage medium of claim 10, wherein the embedded authentication code is decoded and verified as authentic using a learned representation in the authenticator neural network, and wherein to authenticate the content, the authenticator neural network decodes the content created using the authentication code.
A method comprising: generating, by a processor using a generative neural network, samples based on feedback received from a discriminator neural network and from an authenticator neural network, the generative neural network is to trick the discriminator neural network to identify the generated samples as real content samples; digesting, by the processor using the authenticator neural network, the real content samples, the generated samples from the generative neural network, and an authentication code; embedding, by the processor using the authenticator neu-ral network, the authentication code into the generated samples from the generative neural network as an embedded authentication code by contributing to a generator loss provided to the generative neural net-work; generating, by the processor using the generative neural network, content comprising the embedded authentica-tion code; and verifying, by the processor using the authenticator neural network, the content based on the embedded authenti-cation code; wherein a combination of the generative neural network, the discriminator neural network, and the authenticator neural network comprise a generative adversarial net-work (GAN).
The method of claim 16, wherein the generative neural network to generate adversarial loss comprising a first loss function between the generative neural network and the discriminator neural network, and wherein the first loss function is to characterize whether the content from the generative neural network is following a first distribution of original content that is input to the generative neural net-work.
Layer stacks claimed or described, ordered top of device to substrate.
authenticator-integrated GAN apparatus
No layer stack recorded.
non-transitory computer-readable storage medium with authenticator-integrated GAN instructions
No layer stack recorded.
authenticator-integrated GAN method executed by processor
No layer stack recorded.
Patents and literature cited by this patent (applicant and examiner references).
Cited patents · 6
Cited non-patent literature · 4
Cited non-patent literature · 4
Cited non-patent literature · 4
Cited non-patent literature · 4
