-
公开(公告)号:US12288393B2
公开(公告)日:2025-04-29
申请号:US17798969
申请日:2021-06-04
Applicant: Google LLC
Inventor: Rudy Bunel , Jim Huibrecht Winkens , Abhijit Guha Roy , Olaf Ronneberger , Seyed Mohammadali Eslami , Ali Taylan Cemgil , Simon Kohl
Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training a neural network to (i) generate accurate network outputs for a machine learning task and (ii) generate intermediate outputs that can be used to reliably classify out-of-distribution inputs. In one aspect, a method comprises: training the neural network using supervised and contrastive losses, comprising repeatedly performing operations including: obtaining first and second network inputs; processing each network input using the neural network to generate its respective network input embedding; processing the first network input using the neural network to generate a network output; and adjusting the network parameter values using supervised and contrastive loss gradients, wherein: the supervised loss is based on: (i) the network output, and (ii) a corresponding target network output; and the contrastive loss is based on at least: (i) the first network input embedding, and (ii) the second network input embedding.
-
2.
公开(公告)号:US11954902B2
公开(公告)日:2024-04-09
申请号:US17114586
申请日:2020-12-08
Applicant: Google LLC
Inventor: Jeffrey De Fauw , Joseph R. Ledsam , Bernardino Romera-Paredes , Stanislav Nikolov , Nenad Tomasev , Samuel Blackwell , Harry Askham , Xavier Glorot , Balaji Lakshminarayanan , Trevor Back , Mustafa Suleyman , Pearse A. Keane , Olaf Ronneberger , Julien Robert Michel Cornebise
IPC: G06V10/82 , G06F18/21 , G06F18/2413 , G06F18/25 , G06T7/00 , G06T11/00 , G06V10/44 , G06V10/764 , G06V10/80
CPC classification number: G06V10/82 , G06F18/217 , G06F18/24133 , G06F18/254 , G06T7/0012 , G06T11/003 , G06V10/454 , G06V10/764 , G06V10/809 , G06T2207/10101 , G06T2207/20081 , G06T2207/20084 , G06T2207/30041 , G06V2201/03
Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.
-
3.
公开(公告)号:US11676281B2
公开(公告)日:2023-06-13
申请号:US17380914
申请日:2021-07-20
Applicant: Google LLC
Inventor: Stanislav Nikolov , Samuel Blackwell , Jeffrey De Fauw , Bernardino Romera-Paredes , Clemens Ludwig Meyer , Harry Askham , Cian Hughes , Trevor Back , Joseph R. Ledsam , Olaf Ronneberger
CPC classification number: G06T7/11 , A61B5/7267 , A61B6/032 , A61B6/501 , A61N5/1039 , G06N3/08 , G06T7/62 , G06T2207/10081 , G06T2207/20081 , G06T2207/20084 , G06T2207/30196
Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for segmenting a medical image. In one aspect, a method comprises: receiving a medical image that is captured using a medical imaging modality and that depicts a region of tissue in a body; and processing the medical image using a segmentation neural network to generate a segmentation output. The segmentation neural network can include a sequence of multiple encoder blocks and a decoder subnetwork. Training the segmentation neural network can include determining a set of error values for a segmentation channel; identifying the highest error values from the set of error values for the segmentation channel; and determining a segmentation loss based on the highest error values identified for the segmentation channel.
-
4.
公开(公告)号:US20220012891A1
公开(公告)日:2022-01-13
申请号:US17380914
申请日:2021-07-20
Applicant: Google LLC
Inventor: Stanislav Nikolov , Samuel Blackwell , Jeffrey De Fauw , Bernardino Romera-Paredes , Clemens Ludwig Meyer , Harry Askham , Cian Hughes , Trevor Back , Joseph R. Ledsam , Olaf Ronneberger
Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for segmenting a medical image. In one aspect, a method comprises: receiving a medical image that is captured using a medical imaging modality and that depicts a region of tissue in a body; and processing the medical image using a segmentation neural network to generate a segmentation output, wherein the segmentation neural network comprises a sequence of multiple encoder blocks, wherein: each encoder block is a residual neural network block comprising one or more two-dimensional convolutional neural network layers, one or more three-dimensional convolutional neural network layers, or both, and each encoder block is configured to process a respective encoder block input to generate a respective encoder block output wherein a spatial resolution of the encoder block output is lower than a spatial resolution of the encoder block input.
-
5.
公开(公告)号:US11100647B2
公开(公告)日:2021-08-24
申请号:US16565384
申请日:2019-09-09
Applicant: Google LLC
Inventor: Stanislav Nikolov , Samuel Blackwell , Jeffrey De Fauw , Bernardino Romera-Paredes , Clemens Ludwig Meyer , Harry Askham , Cian Hughes , Trevor Back , Joseph R. Ledsam , Olaf Ronneberger
Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for segmenting a medical image. In one aspect, a method comprises: receiving a medical image that is captured using a medical imaging modality and that depicts a region of tissue in a body; and processing the medical image using a segmentation neural network to generate a segmentation output, wherein the segmentation neural network comprises a sequence of multiple encoder blocks, wherein: each encoder block is a residual neural network block comprising one or more two-dimensional convolutional neural network layers, one or more three-dimensional convolutional neural network layers, or both, and each encoder block is configured to process a respective encoder block input to generate a respective encoder block output wherein a spatial resolution of the encoder block output is lower than a spatial resolution of the encoder block input.
-
6.
公开(公告)号:US20210118198A1
公开(公告)日:2021-04-22
申请号:US17114586
申请日:2020-12-08
Applicant: Google LLC
Inventor: Jeffrey De Fauw , Joseph R. Ledsam , Bernardino Romera-Paredes , Stanislav Nikolov , Nenad Tomasev , Samuel Blackwell , Harry Askham , Xavier Glorot , Balaji Lakshminarayanan , Trevor Back , Mustafa Suleyman , Pearse A. Keane , Olaf Ronneberger , Julien Robert Michel Cornebise
Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a final classification output for an image of eye tissue. The image is provided as input to each of one or more segmentation neural networks to obtain one or more segmentation maps of the eye tissue in the image. A respective classification input is generated from each of the segmentation maps. For each of the segmentation maps, the classification input for the segmentation map is provided as input to each of one or more classification neural networks to obtain, for each segmentation map, a respective classification output from each classification neural network. A final classification output for the image is generated from the respective classification outputs for each of the segmentation maps.
-
-
-
-
-