公开(公告)号：US11586988B2

公开(公告)日：2023-02-21

申请号：US16438664

申请日：2019-06-12

Applicant: FUJITSU LIMITED

Inventor： Mengjiao Wang ,  Rujie Liu

IPC: G06N20/20 ,  G06K9/62

Abstract: There are provided a method of knowledge transferring, an information processing apparatus and a storage medium. The method of knowledge transferring includes: obtaining a first model which has been trained in advance with respect to a predetermined task; and training a second model with respect to the predetermined task by utilizing a comprehensive loss function, such that the second model has knowledge of the first model. The comprehensive loss function is based on a first loss function weighted by accuracy of an output result of the first model for a training sample in regard to the predetermined task, and a second loss function. The first loss function represents a difference between processing results of the second model and the first model for the training sample. The second loss function represents accuracy of an output result of the second model for the training sample in regard to the predetermined task.

公开(公告)号：US11270139B2

公开(公告)日：2022-03-08

申请号：US16737370

申请日：2020-01-08

Applicant: FUJITSU LIMITED

Inventor： Meng Zhang ,  Rujie Liu

IPC: G06K9/00 ,  G06K9/62

Abstract: An apparatus for training a classifying model comprises: a first obtaining unit configured to input a sample image to a first machine learning framework, to obtain a first classification probability and a first classification loss; a second obtaining unit configured to input a second image to a second machine learning framework, to obtain a second classification probability and a second classification loss, the two machine learning frameworks having identical structures and sharing identical parameters; a similarity loss calculating unit configured to calculate a similarity loss related to a similarity between the first classification probability and the second classification probability; a total loss calculating unit configured to calculate the sum of the similarity loss, the first classification loss and the second classification loss, as a total loss; and a training unit configured to adjust parameters of the two machine learning frameworks to obtain a trained classifying model.

公开(公告)号：US11200464B2

公开(公告)日：2021-12-14

申请号：US16744220

申请日：2020-01-16

Applicant: FUJITSU LIMITED

Inventor： Wei Shen ,  Rujie Liu

IPC: G06K9/62 ,  G06N3/04

Abstract: An information processing method includes: inputting sample image into a machine learning architecture to obtain a first feature, and causing a first classifier to calculate a first classification loss; calculating a second feature based on the first feature and a predetermined first mask, and inputting the second feature into the first classifier to calculate an entropy loss; calculating a second mask based on the first mask and the entropy loss to maximize the entropy loss; obtaining an adversarial feature based on the first feature and the second mask, where the adversarial feature is complementary to the second feature; causing, by training the first classifier and the second classifier in association with each other, the second classifier to calculate a second classification loss based on the adversarial feature; and adjusting parameters of the machine learning architecture, the first classifier and the second classifier, to obtain a trained machine learning architecture.

公开(公告)号：US11113513B2

公开(公告)日：2021-09-07

申请号：US16738393

申请日：2020-01-09

Applicant: FUJITSU LIMITED

Inventor： Meng Zhang ,  Rujie Liu

IPC: G06K9/46 ,  G06K9/00 ,  G06K9/62 ,  G06N3/08

Abstract: An apparatus and method for training a classification model and an apparatus for classifying with a classification model are disclosed. The apparatus for training a classification model comprises: a local area obtainment unit to, obtain predetermined local area of each sample image; a feature extraction unit to, with respect to each sample image, set corresponding numbers of feature extraction layers for the global area and each predetermined local area, to extract a global feature of the global area and a local feature of each predetermined local area, wherein the global area and the predetermined local areas share at least one feature extraction layer in which the global feature and each local feature are combined; and a loss determination unit to calculate a loss function of the sample image based on combined features of each sample image, and to train the classification model based on the loss function.

公开(公告)号：US20210142150A1

公开(公告)日：2021-05-13

申请号：US17090032

申请日：2020-11-05

Applicant: FUJITSU LIMITED

Inventor： Meng Zhang ,  Fei Li ,  Rujie Liu

IPC: G06N3/04 ,  G06K9/62

Abstract: An information processing device and method, and a device for classifying with a model are provided. The information processing device includes a first training unit being configured to train a first model using a first training sample set, to obtain a trained first model; a second training unit being configured to train the trained first model using a second training sample set while maintaining a predetermined portion of characteristics of the trained first model, to obtain a trained second model, and a third training unit being configured to train a third model using the second training sample set while causing a difference between classification performances of the trained second model and the third model to be within a first predetermined range, to obtain a trained third model as a final model.

公开(公告)号：US20190138854A1

公开(公告)日：2019-05-09

申请号：US16179292

申请日：2018-11-02

Applicant: Fujitsu Limited

Inventor： Meng Zhang ,  Rujie Liu ,  Jun Sun

IPC: G06K9/62 ,  G06K9/00

CPC classification number: G06K9/6262 ,  G06K9/00281 ,  G06K9/00288 ,  G06K9/6256

Abstract: A method and apparatus for removing black eyepits and sunglasses in first actual scenario data having an image containing a face acquired from an actual scenario, to obtain second actual scenario data; counting a proportion of wearing glasses in the second actual scenario data; dividing original training data composed of an image containing a face into wearing-glasses and not-wearing-glasses first and second training data, where a proportion of wearing glasses in the original training data is lower than a proportion in the second actual scenario data; generating wearing-glasses third training data based on glasses data and the second training data; generating fourth training data in which a proportion of wearing glasses is equal to the proportion of wearing glasses in the second actual scenario data, based on the third training data and the original training data; and training a face recognition model based on the fourth training data.

公开(公告)号：US09830380B2

公开(公告)日：2017-11-28

申请号：US14307687

申请日：2014-06-18

Applicant: Fujitsu Limited

Inventor： Xi Liu ,  Rujie Liu

IPC: G06F17/30

CPC classification number: G06F17/30663 ,  G06F17/30017 ,  G06F17/30247 ,  G06F17/30256 ,  G06F17/30259 ,  G06F17/30268 ,  G06F17/3028

Abstract: Embodiments provide a multimodality-based image tagging apparatus and a method for the same. The image tagging apparatus includes: a score generating unit configured to generate, for an inquiry image, multiple groups of first scores about all tags in an tagging dictionary by using a training image and multiple modalities of an image; a late-fusion unit configured to fuse the obtained multiple groups of scores to obtain final scores about all the tags; and a tag selecting unit configured to select one or more tag(s) with relatively large tag scores as tag(s) of the inquiry image according to the final scores about all the tags. With the embodiments, multiple modalities may be effectively fused, and a more robust and accurate image tagging result may be obtained.

公开(公告)号：US20170294191A1

公开(公告)日：2017-10-12

申请号：US15477687

申请日：2017-04-03

Applicant: FUJITSU LIMITED

Inventor： Ziqiang SHI ,  Liu LIU ,  Rujie Liu

IPC: G10L17/02 ,  G10L15/30 ,  G10L17/10 ,  G10L15/22 ,  G10L15/20 ,  G10L17/04

CPC classification number: G10L17/02 ,  G10L15/20 ,  G10L15/22 ,  G10L15/30 ,  G10L17/04 ,  G10L17/10 ,  G10L2015/227

Abstract: The present invention discloses a method for speaker recognition and an apparatus for speaker recognition. The method for speaker recognition comprises: extracting, from a speaker-to-be-recognized corpus, voice characteristics of a speaker to be recognized: obtaining a speaker-to-be-recognized model based on the extracted voice characteristics of the speaker to be recognized, a universal background model UBM reflecting distribution of the voice characteristics in a characteristic space, a gradient universal speaker model GUSM reflecting statistic values of changes of the distribution of the voice characterizes in the characteristic space and a total change matrix reflecting environmental changes; and comparing the speaker-to-be-recognized model with known speaker models, to determine whether or not the speaker to be recognized is one of known speakers.

公开(公告)号：US09466010B2

公开(公告)日：2016-10-11

申请号：US14304324

申请日：2014-06-13

Applicant: FUJITSU LIMITED

Inventor： Qiong Cao ,  Rujie Liu

IPC: G06K9/62 ,  G06K9/52

CPC classification number: G06K9/6211 ,  G06K9/52

Abstract: An image similarity determining device and method and an image feature acquiring device and method are provided. The image similarity determining device comprises a preprocessing unit for extracting feature points of each input image region of an input image and each image region to be matched of a data source image; a matched feature point set determining unit for determining one to one matched feature point pairs between input image regions and image regions to be matched to determine matched feature point sets; a geometry similarity determining unit for determining a geometry similarity between the input image region and the image region to be matched based on distribution of respective feature points in the matched feature point sets; and an image similarity determining unit for determining similarity between input image and data source image based on geometry similarities between input image regions and corresponding image regions to be matched.

Abstract translation: 提供了图像相似性确定装置和方法以及图像特征获取装置和方法。 图像相似性确定装置包括：预处理单元，用于提取输入图像的每个输入图像区域的特征点和要与数据源图像匹配的每个图像区域; 匹配特征点集确定单元，用于确定输入图像区域和要匹配的图像区域之间的一对匹配的特征点对，以确定匹配的特征点集合; 几何相似度确定单元，用于基于匹配的特征点集合中的各个特征点的分布来确定输入图像区域和要匹配的图像区域之间的几何相似度; 以及图像相似性确定单元，用于基于输入图像区域和要匹配的对应图像区域之间的几何相似度来确定输入图像和数据源图像之间的相似性。

公开(公告)号：US11514272B2

公开(公告)日：2022-11-29

申请号：US16736180

申请日：2020-01-07

Applicant: FUJITSU LIMITED

Inventor： Meng Zhang ,  Rujie Liu

IPC: G06K9/62 ,  G06N20/00 ,  G06N3/08

Abstract: An apparatus for training a classification model includes: a feature extraction unit configured to set, with respect to each training set of a first predetermined number of training sets, feature extraction layers, and extract features of a sample image, where at least two of the training sets at least partially overlap; a feature fusion unit configured to set, with respect to training set, feature fusion layers, and perform a fusion on the extracted features of the sample image; and a loss determination unit configured to set, with respect to training set, a loss determination layer, calculate a loss function of the sample image based on the fused feature of the sample image, and train a classification model based on the loss function. The first predetermined number of training sets share at least one layer of feature fusion layers and feature extraction layers set with respect to each training set.