公开(公告)号：US11443162B2

公开(公告)日：2022-09-13

申请号：US16549715

申请日：2019-08-23

Applicant: Google LLC

Inventor： Ming-Hsuan Yang ,  Xiaojie Jin ,  Joshua Foster Slocum ,  Shengyang Dai ,  Jiang Wang

IPC: G06N3/04 ,  G06N20/00 ,  G06F16/901

Abstract: Methods, and systems, including computer programs encoded on computer storage media for neural network architecture search. A method includes defining a neural network computational cell, the computational cell including a directed graph of nodes representing respective neural network latent representations and edges representing respective operations that transform a respective neural network latent representation; replacing each operation that transforms a respective neural network latent representation with a respective linear combination of candidate operations, where each candidate operation in a respective linear combination has a respective mixing weight that is parameterized by one or more computational cell hyper parameters; iteratively adjusting values of the computational cell hyper parameters and weights to optimize a validation loss function subject to computational resource constraints; and generating a neural network for performing a machine learning task using the defined computational cell and the adjusted values of the computational cell hyper parameters and weights.

公开(公告)号：US20210056378A1

公开(公告)日：2021-02-25

申请号：US16549715

申请日：2019-08-23

Applicant: Google LLC

Inventor： Ming-Hsuan Yang ,  Xiaojie Jin ,  Joshua Foster Slocum ,  Shengyang Dai ,  Jiang Wang

IPC: G06N3/04 ,  G06N20/00 ,  G06F16/901

Abstract: Methods, and systems, including computer programs encoded on computer storage media for neural network architecture search. A method includes defining a neural network computational cell, the computational cell including a directed graph of nodes representing respective neural network latent representations and edges representing respective operations that transform a respective neural network latent representation; replacing each operation that transforms a respective neural network latent representation with a respective linear combination of candidate operations, where each candidate operation in a respective linear combination has a respective mixing weight that is parameterized by one or more computational cell hyper parameters; iteratively adjusting values of the computational cell hyper parameters and weights to optimize a validation loss function subject to computational resource constraints; and generating a neural network for performing a machine learning task using the defined computational cell and the adjusted values of the computational cell hyper parameters and weights.

公开(公告)号：US20220414425A1

公开(公告)日：2022-12-29

申请号：US17821076

申请日：2022-08-19

Applicant: Google LLC

Inventor： Ming-Hsuan Yang ,  Xiaojie Jin ,  Joshua Foster Slocum ,  Shengyang Dai ,  Jiang Wang

IPC: G06N3/04 ,  G06N20/00 ,  G06F16/901

Abstract: Methods, and systems, including computer programs encoded on computer storage media for neural network architecture search. A method includes defining a neural network computational cell, the computational cell including a directed graph of nodes representing respective neural network latent representations and edges representing respective operations that transform a respective neural network latent representation; replacing each operation that transforms a respective neural network latent representation with a respective linear combination of candidate operations, where each candidate operation in a respective linear combination has a respective mixing weight that is parameterized by one or more computational cell hyper parameters; iteratively adjusting values of the computational cell hyper parameters and weights to optimize a validation loss function subject to computational resource constraints; and generating a neural network for performing a machine learning task using the defined computational cell and the adjusted values of the computational cell hyper parameters and weights.