Deep feature generative adversarial neural networks
AssigneeSnap Inc.
InventorsSergey Demyanov, Aleksei Podkin, Aleksei Stoliar, Vadim Velicodnii, Fedor Zhdanov
Life of this patent
Estimating the term…
Abstract
A data item is identified on a device. A neural network that includes an adversarial transformation subnetwork is applied to the data item to generate a modified data item. Output indicative of the modified data item is caused to be presented on the device. The neural network further comprises an encoder and a decoder. The neural network is trained in at least two stages. At least one of the encoder and the decoder is trained in a first stage and the adversarial transformation subnetwork is trained in a second stage.
Claims20
A method comprising:
- identifying a data item on a device;
- applying a neural network including an adversarial transformation subnetwork to apply a transformation to the data item to generate a modified data item, the neural network further comprising an encoder and a decoder, the neural network being trained in at least two stages, at least one of the encoder and the decoder being trained in a first stage and the adversarial transformation subnetwork being trained in a second stage that occurs after the first stage, the encoder and the decoder being fixed during the second stage;
- and
- causing presentation of output indicative of the modified data item on the device.
A system comprising:
- a memory that stores instructions;
- and
- one or more processors configured by the instructions to perform operations comprising:
- identifying a data item on a device;
- applying a neural network including an adversarial transformation subnetwork to apply a transformation to the data item to generate a modified data item, the neural network further comprising an encoder and a decoder, the neural network being trained in at least two stages, at least one of the encoder and the decoder being trained in a first stage and the adversarial transformation subnetwork being trained in a second stage that occurs after the first stage, the encoder and the decoder being fixed during the second stage;
- and
- causing presentation of output indicative of the modified data item on the device.
A non-transitory computer-readable storage medium, the computer-readable storage medium including instructions that when executed by at least one processor, cause the at least one processor to perform operations comprising:
- identifying a data item on a device;
- applying a neural network including an adversarial transformation subnetwork to apply a transformation to the data item to generate a modified data item, the neural network further comprising an encoder and a decoder, the neural network being trained in at least two stages, at least one of the encoder and the decoder being trained in a first stage and the adversarial transformation subnetwork being trained in a second stage that occurs after the first stage, the encoder and the decoder being fixed during the second stage;
- and
- causing presentation of output indicative of the modified data item on the device.
Dependencies are inferred from the claim text.