公开(公告)号：US11636283B2

公开(公告)日：2023-04-25

申请号：US16889125

申请日：2020-06-01

Applicant: DeepMind Technologies Limited

Inventor： Benjamin Poole ,  Aaron Gerard Antonius van den Oord ,  Ali Razavi-Nematollahi ,  Oriol Vinyals

IPC: G06K9/62 ,  G06N3/04 ,  G06N3/08 ,  G06N7/00 ,  G06N3/084

Abstract: A variational autoencoder (VAE) neural network system, comprising an encoder neural network to encode an input data item to define a posterior distribution for a set of latent variables, and a decoder neural network to generate an output data item representing values of a set of latent variables sampled from the posterior distribution. The system is configured for training with an objective function including a term dependent on a difference between the posterior distribution and a prior distribution. The prior and posterior distributions are arranged so that they cannot be matched to one another. The VAE system may be used for compressing and decompressing data.

公开(公告)号：US10671889B2

公开(公告)日：2020-06-02

申请号：US16586014

申请日：2019-09-27

Applicant: DeepMind Technologies Limited

Inventor： Benjamin Poole ,  Aaron Gerard Antonius van den Oord ,  Ali Razavi-Nematollahi ,  Oriol Vinyals

IPC: G06K9/62 ,  G06N3/08 ,  G06N3/04 ,  G06N7/00

Abstract: A variational autoencoder (VAE) neural network system, comprising an encoder neural network to encode an input data item to define a posterior distribution for a set of latent variables, and a decoder neural network to generate an output data item representing values of a set of latent variables sampled from the posterior distribution. The system is configured for training with an objective function including a term dependent on a difference between the posterior distribution and a prior distribution. The prior and posterior distributions are arranged so that they cannot be matched to one another. The VAE system may be used for compressing and decompressing data.

公开(公告)号：US20200104640A1

公开(公告)日：2020-04-02

申请号：US16586014

申请日：2019-09-27

Applicant: DeepMind Technologies Limited

Inventor： Benjamin Poole ,  Aaron Gerard Antonius van den Oord ,  Ali Razavi-Nematollahi ,  Oriol Vinyals

IPC: G06K9/62 ,  G06N7/00 ,  G06N3/04 ,  G06N3/08

Abstract: A variational autoencoder (VAE) neural network system, comprising an encoder neural network to encode an input data item to define a posterior distribution for a set of latent variables, and a decoder neural network to generate an output data item representing values of a set of latent variables sampled from the posterior distribution. The system is configured for training with an objective function including a term dependent on a difference between the posterior distribution and a prior distribution. The prior and posterior distributions are arranged so that they cannot be matched to one another. The VAE system may be used for compressing and decompressing data.