公开(公告)号：US11631025B2

公开(公告)日：2023-04-18

申请号：US15536783

申请日：2016-01-05

申请人： NEC Corporation

发明人： Atsushi Sato

IPC分类号： G06N20/00 ,  G06F40/242

摘要： A learning apparatus includes: an update unit which updates a dictionary used by a classifier; a calculation unit which calculates, by using a dictionary updated and one or more samples with labeling being samples assigned with labels, a ratio to a number of the samples with labeling as a loss with respect to all the samples with labeling; and a determination unit which determines whether to update the dictionary, by using the loss, wherein, when the determination unit determines to update the dictionary, the update unit updates the dictionary by using the samples with labeling added with a new sample with labeling, and wherein the determination unit determines whether to update the dictionary, by using a loss calculated by using the updated dictionary and a loss calculated by using the dictionary before updating with respect to all the samples with labeling before adding the new sample with labeling.

公开(公告)号：US11281686B2

公开(公告)日：2022-03-22

申请号：US15734442

申请日：2018-06-04

申请人： NEC Corporation

发明人： Devendra Dhaka ,  Masato Ishii ,  Atsushi Sato

IPC分类号： G06F16/00 ,  G06F16/25 ,  G06F16/2455 ,  G06F16/28 ,  G06F16/21

摘要： The information processing apparatus (2000) of the example embodiment 1 includes an acquisition unit (2020), a clustering unit (2040), a transformation unit (2060) and modeling unit (2080). Until a predetermined termination condition is determined, the clustering unit (2040) repeatedly preforms: 1) optimizing the posterior parameters for clustering assignment for each data streams; 2) optimizes the posterior parameters for each determined cluster and for each time frame; 3) optimizes the posterior parameters for individual responses for each data stream; 4) optimizes the posterior parameters for latent states, via approximating the observation model through non-conjugate inference. The transformation unit (2060) transforms the latent states into parameters of the observation model, through a transformation function. The modeling unit (2060) generates the model data, which including all the optimized parameters of all the model latent variables, optimized inside the clustering unit (2040).

公开(公告)号：US10657672B2

公开(公告)日：2020-05-19

申请号：US15763863

申请日：2015-09-30

申请人： NEC Corporation

发明人： Kapik Lee ,  Takashi Shibata ,  Atsushi Sato

IPC分类号： G06T7/73

摘要： The provided is an image processing technology to avoid inaccuracy of inferring a position of an object due to wrong model assumption.An image processing device according to an exemplary aspect of the present invention includes: a matching unit 106 that estimates similarity of input image to particle images including areas of particles at relative positions, true positions of the particles in the particle images being given; and a calculation unit 108 that calculates positions of particles in the input image based on the similarity and the true positions.

公开(公告)号：US10225575B2

公开(公告)日：2019-03-05

申请号：US15506009

申请日：2014-09-03

申请人： NEC Corporation

发明人： Lee Kapik ,  Atsushi Sato ,  Takashi Shibata

IPC分类号： H04N19/59 ,  H04N5/21 ,  H04N7/01 ,  G06T3/40 ,  H04N19/14 ,  H04N19/172 ,  H04N19/54

摘要： An image processing device according to one of the exemplary aspects of the present invention includes: inferring means for selecting, for each of local unknown patches including a target unknown patch, candidate patches from a plurality of input patches based on similarity to the local unknown patch, the local unknown patches being images generated from a part of an unknown image, the plurality of input patches being images generated from a plurality of input images, a subject ID (Identifier) being correlated with the input patches that are generated from an input image to which the subject ID is assigned in the plurality of input image; first score calculation means for calculating a score representing nearness of a candidate patch in the candidate patches to a local unknown patch in the local unknown patches; and patch replacement means for calculating a score summation for the subject ID by summing up scores of the candidate patches being correlated with a same subject ID in the candidate patches of the local unknown patches, and selecting, as a selected patch being used for reconstruction of a reconstruction image, a candidate patch that is correlated with the subject ID for which the score summation is highest from the candidate patches selected for the target unknown patch.

公开(公告)号：US20170280158A1

公开(公告)日：2017-09-28

申请号：US15506009

申请日：2014-09-03

申请人： NEC Corporation

发明人： Lee Kapik ,  Atsushi Sato ,  Takashi Shibata

IPC分类号： H04N19/59 ,  H04N19/54 ,  H04N19/14 ,  H04N19/172

CPC分类号： H04N19/59 ,  G06T3/4053 ,  H04N5/21 ,  H04N7/0125 ,  H04N19/14 ,  H04N19/172 ,  H04N19/54

摘要： The present invention aims to provide an image processing device which enables reconstruction of a sharp high resolution image similar to the original of the unknown input image.An image processing device according to one of the exemplary aspects of the present invention includes: inferring means for selecting, for each of local unknown patches including a target unknown patch, candidate patches from a plurality of input patches based on similarity to the local unknown patch, the local unknown patches being images generated from a part of an unknown image, the plurality of input patches being images generated from a plurality of input images, a subject ID (Identifier) being correlated with the input patches that are generated from a input image to which the subject ID is assigned in the plurality of input image; first score calculation means for calculating a score representing nearness of a candidate patch in the candidate patches to a local unknown patch in the local unknown patches; and patch replacement means for calculating a score summation for the subject ID by summing up scores of the candidate patches being correlated with a same subject ID in the candidate patches of the local unknown patches, and selecting, as a selected patch being used for reconstruction of a reconstruction image, a candidate patch that is correlated with the subject ID for which the score summation is highest from the candidate patches selected for the target unknown patch.

公开(公告)号：US11195299B2

公开(公告)日：2021-12-07

申请号：US16644217

申请日：2017-09-12

申请人： NEC Corporation

发明人： Kapik Lee ,  Takashi Shibata ,  Atsushi Sato

IPC分类号： G06T7/00 ,  G06T7/73 ,  G06K9/00 ,  G06K9/62

摘要： An image processing device according to an aspect of the present invention includes: at least one memory storing a set of instructions; and at least one processor configured to execute the set of instructions to: extract a cluster image including a cluster of particles from a particle image; estimate particle density of the cluster image; determine numbers based on the particle density, and synthesize a second dictionary image from first dictionary images having densities each being equal to the numbers, the first dictionary images being images of particles of which positions is obtained; match the second dictionary image with the partial image, and determine positions of particles in the cluster image based on a result of the matching and the positions of particles of the first dictionary images; and output the determined positions of particles.

公开(公告)号：US11537930B2

公开(公告)日：2022-12-27

申请号：US14770889

申请日：2013-11-01

申请人： NEC Corporation

发明人： Atsushi Sato

IPC分类号： G06N20/00

摘要： An information processing device which performs semi-supervised learning is provided with: a dictionary input circuit for acquiring a dictionary, a parameter group used by an identification device; a boundary determination circuit which obtains an identification boundary for the dictionary on the basis of the dictionary, supervised data, and labelled unsupervised data; a labelling circuit which labels the unsupervised data in accordance with the identification boundary; a loss calculation circuit which calculates the sum total of supervised-data loss calculated in accordance with the labels assigned in advance and the labels based on the identification boundary, and unsupervised-data loss calculated such that further from the identification boundary the smaller the loss; a dictionary update circuit which updates the dictionary such that the sum-total loss is reduced; and a dictionary output circuit which outputs the updated dictionary.