-
公开(公告)号：US20210192331A1
公开(公告)日：2021-06-24
申请号：US16078121
申请日：2017-02-14
Applicant: NEC CORPORATION
Inventor： Masato ISHII
Abstract: A neural network training device according to an exemplary aspect of the present invention includes: a memory that stores a set of instructions; and at least one central processing unit (CPU) configured to execute the set of instructions to: determine a regularization strength for each layer, based on an initialized network; and train a network, based on the initialized network and the determined regularization strength, wherein the at least one CPU is further configured to determine the regularization strength in such a way that a difference between magnitude of a parameter update amount calculated from a loss function and magnitude of a parameter update amount calculated from a regularization term falls within a predetermined range.
-
公开(公告)号：US20210019636A1
公开(公告)日：2021-01-21
申请号：US17043309
申请日：2018-05-11
Applicant: NEC Corporation
Inventor： Masato ISHII , Takashi TAKENOUCHI , Masashi SUGIYAMA
Abstract: This prediction model preparation device is provided with: a calculation means which calculates, from a datum in which a sample and a label are associated with each other, an importance level according to the difference between a first possibility that an event influencing the sample occurs in a source domain and a second possibility that the event occurs in a target domain; and a preparation means which constructs prepares a prediction model relating to the target domain by calculating association between the sample and the label included in the datum to which the importance level is added.
-
Patent Agency Ranking
13.发明申请公开(公告)号：US20150371098A1
公开(公告)日：2015-12-24
申请号：US14769532
申请日：2014-02-05
Applicant: NEC CORPORATION
Inventor： Masato ISHII
CPC classification number: G06K9/00892 , G06F21/32
Abstract: The purpose of the present invention is to prevent a reduction in authentication accuracy caused by identity fraud. A score calculation unit compares each of a plurality of types of biological information, acquired as acquired biological information from a target person of identity verification, to the same type of registered biological information registered in advance. Based on the comparison, the score calculation unit calculates an authentication score that expresses the degree of similarity between the acquired biological information and the registered information, for each type of acquired biological information. For each type of the acquired biological information, a probability calculation unit calculates as an identity fraud probability using the calculated authentication score to. A determination unit determines whether the target person of identity verification is the registered person, and/or determines whether the target person of identity verification is fraudulently pretending to be the registered person.
Abstract translation: 本发明的目的是防止由身份欺诈引起的认证精度降低。 得分计算单元将从身份验证的目标人获取的获取的生物信息的多种生物信息中的每一种与预先登记的相同类型的登记生物信息进行比较。 基于比较，得分计算单元计算表示所获取的生物信息与登记信息之间的相似度的认证分数，对于每种获得的生物信息。 对于每种类型的获取的生物信息，概率计算单元使用计算的认证得分计算为身份欺诈概率。 确定单元确定身份验证的目标人员是否是注册人，和/或确定身份验证的目标人员是否欺诈地假装为注册人。
-
公开(公告)号：US20220058313A1
公开(公告)日：2022-02-24
申请号：US17413200
申请日：2018-12-25
Applicant: NEC Corporation
Inventor： Devendra DHAKA , Masato ISHII , Atsushi SATO
Abstract: The information processing apparatus (2000) of the example embodiment 1 includes an acquisition unit (2020), a modeling unit (2040), an output unit (2080). The acquisition unit (2020) acquires a plurality of trajectory data. The trajectory data represents a time-sequence of observed positions of an object. The modeling unit (2040) assigns one of groups for each trajectory data. The modeling unit (2040) generates a generative model for each group. The generative model represents trajectories assigned to the corresponding group by a common time-sequence of velocity transformations. The velocity transformation represents a transformation of velocity of the object from a previous time frame, and is represented using a set of motion primitives defined in common for all groups. The output unit (2060) outputs the generated generative models.
-
公开(公告)号：US20210133552A1
公开(公告)日：2021-05-06
申请号：US16477234
申请日：2018-01-17
Applicant: NEC Corporation
Inventor： Masato ISHII
Abstract: A neural network learning device 20 is equipped with: a determination module 22 that determines the size of a local region in learning information 200 which is to be learned by a neural network 21 containing multiple layers, said determination being made for each layer, on the basis of the structure of the neural network 21; and a control module 25 that, on the basis of size of the local region as determined by the determination module 22, extracts the local region from the learning information 200, and performs control such that the learning of the learning information represented by the extracted local region by the neural network 200 is carried out repeatedly while changing the size of the extracted local region, and thus, a reduction in the generalization performance of the neural network can be avoided even when there is little learning data.
-
公开(公告)号：US20200272897A1
公开(公告)日：2020-08-27
申请号：US16762571
申请日：2018-11-19
Applicant: NEC CORPORATION
Inventor： Masato ISHII
Abstract: A learning device according to the present invention includes, in semi-supervised learning using domain information: a memory; and a processor. The processor performs operations. The operations includes: including a first neural network outputting data after predetermined conversion by using first data including the domain information and second data not including the domain information, a second neural network outputting a result of predetermined processing by using data after the conversion, and a third neural network outputting a result of domain discrimination by using data after the conversion; calculating a first loss being a loss of the domain discrimination; calculating a second loss being an unsupervised loss; calculating a third loss in the predetermined processing; and modifying a parameter of each of the first neural network to the third neural network in such a way as to decrease the second loss and the third loss and increase the first loss.
-
公开(公告)号：US20150363709A1
公开(公告)日：2015-12-17
申请号：US14763702
申请日：2013-08-26
Applicant: NEC CORPORATION
Inventor： Toshio KAMEI , Masato ISHII
CPC classification number: G06N20/00 , G06F7/72 , G06K9/6269 , G06K9/6276
Abstract: A classifier learning apparatus (100) includes: an object acquisition unit (101) that acquires a set of reference vectors and assigned category information of the respective reference vectors as a processing object; a specifying unit (102) that specifies an internal nearest neighbor reference vector nearest to a sample vector among the reference vectors assigned to the same category as the sample vector and specifies an external nearest neighbor reference vector nearest to the sample vector among the reference vectors assigned to a category different from that of the sample vector; a calculation unit (103) that calculates an evaluation value of the processing object using a distance between the sample vector and a classification boundary formed by the internal nearest neighbor reference vector and the external nearest neighbor reference vector; and an updating unit (104) that updates an original set of reference vectors and original assigned category information with the processing object based on the evaluation value.
Abstract translation: 分类器学习装置（100）包括：对象获取单元（101），其获取一组参考矢量，并将各个参考矢量的分配类别信息作为处理对象; 指定单元（102），其指定与分配给与样本矢量相同类别的参考向量中最接近样本矢量的内部最近邻参考矢量，并指定分配给参考矢量中最接近样本矢量的外部最近邻参考矢量 到与样本载体不同的类别; 计算单元，其使用所述采样矢量与由所述内部最近邻参考矢量和所述外部最近邻参考矢量形成的分类边界之间的距离来计算所述处理对象的评价值; 以及更新单元（104），其基于所述评估值，利用所述处理对象来更新原始参考矢量集合和原始分配的类别信息。
-
-
-
-
-
-
Search Results
17
records
(took 0.013 seconds)
