公开(公告)号：US20160117588A1

公开(公告)日：2016-04-28

申请号：US14895023

申请日：2014-06-06

Applicant: NEC CORPORATION

Inventor： Yusuke MURAOKA ,  Yukitaka KUSUMURA ,  Hironori MIZUGUCHI ,  Dai KUSUI

IPC: G06N5/02 ,  G06N99/00

CPC classification number: G06N5/02 ,  G06N20/00 ,  G06Q10/06375

Abstract: The prediction function creation device according to the present invention for creating a prediction function to derive an objective variable by using a set of samples that include explanatory variables and an objective variable, the device includes: a clustering unit that clusters the respective samples by giving labels, and assigns weights to each label in accordance with patterns of missing values for the explanatory variables in labeled samples; a child model creation unit that makes portions of the training data partial training data on the basis of the weights, and determines an explanatory variable that constitutes the prediction function on the basis of patterns of missing values for the explanatory variables in the samples; and a mixture model creation unit that creates the prediction function with respect to each pattern of missing values by using the explanatory variable and the determined partial training data.

Abstract translation: 根据本发明的预测函数创建装置，用于通过使用包括解释变量和目标变量的一组样本来创建预测函数以导出目标变量，所述设备包括：聚类单元，通过给出标签来聚集各个样本 ，并根据标记样本中的解释变量的缺失值模式为每个标签分配权重; 子模型创建单元，其基于权重构成训练数据部分训练数据的部分，并且基于样本中的解释变量的缺失值的模式来确定构成预测函数的解释变量; 以及混合模型创建单元，其通过使用所述解释变量和所确定的部分训练数据来创建关于缺失值的每个模式的预测函数。

公开(公告)号：US20200057948A1

公开(公告)日：2020-02-20

申请号：US16346004

申请日：2017-10-05

Applicant: NEC CORPORATION

Inventor： Ryohei FUJIMAKI ,  Yukitaka KUSUMURA ,  Masato ASAHARA ,  Yusuke MURAOKA

IPC: G06N5/04 ,  G06F16/22 ,  G06N20/00

Abstract: A feature design unit 81 designs, from relational data, a feature as a variable likely to affect an objective variable. A feature generating unit 82 generates the designed feature, from the relational data. A learning unit 83 learns a prediction model, on the basis of the generated feature.

公开(公告)号：US20180373764A1

公开(公告)日：2018-12-27

申请号：US15774645

申请日：2016-11-14

Applicant: NEC Corporation

Inventor： Yukitaka KUSUMURA ,  Ryohei FUJIMAKI

IPC: G06F17/30

Abstract: A table storage unit 81 stores a first table including an objective variable and a second table different in granularity from the first table. A descriptor creation unit 82 creates a feature descriptor for generating a feature which is a variable that can influence the objective variable, from the first table and the second table. The descriptor creation unit 82 creates a plurality of feature descriptors, each by generating a combination of a mapping condition element indicating a mapping condition for rows in the first table and the second table and a reduction method element indicating a reduction method for reducing, for each objective variable, data of each column included in the second table.

公开(公告)号：US20170109629A1

公开(公告)日：2017-04-20

申请号：US15316075

申请日：2015-02-13

Applicant: NEC Corporation

Inventor： Yukitaka KUSUMURA

IPC: G06N3/12 ,  G06N99/00

Abstract: An enumeration plan generation unit 81 generates a set of logical formula structures each representing a way of combining logical formula expressions each representing a combination of features by use of the features of learning data items and the maximum number of features to be combined, and generates partial logical formula structures by dividing a logical formula expression included in each of the generated logical formula structures into two, and generates an enumeration plan in which the partial logical formula structures are linked to the logical formula structure from which the partial logical formula structures are divided. The feature generation unit 82 generates a new feature that is a combination of the features corresponding to the generated partial logical formula structures. Furthermore, the enumeration plan generation unit 81 divides the logical formula structure into two such that the numbers of the features included in the two partial logical formula structures generated from each of the logical formula structures are substantially equal.

公开(公告)号：US20180136076A1

公开(公告)日：2018-05-17

申请号：US15573039

申请日：2016-03-10

Applicant: NEC CORPORATION

Inventor： Yukitaka KUSUMURA ,  Sergey TARASENKO ,  Riki ETO ,  Yusuke MURAOKA ,  Ryohei FUJIMAKI

IPC: G01M3/28

CPC classification number: G01M3/28

Abstract: This invention provides a water-leakage state estimation system configured to estimate a state of a water leakage in a specific area of a water distribution network. A learning unit is configured to: receive labeled data, which is labeled so as to separate past flow rate data into abnormal values and normal values, and past environment state condition data; build a prediction model for predicting the normal values in the labeled data through learning; and determine a score parameter defining a length of a period involving data to be verified through learning as well. A water-leakage estimation unit is configured to: compare predicted flow rate data obtained by supplying current environment condition data into the prediction model and current flow rate data to produce error values; and calculate an average value of the error values in the period of a window width defined by the score parameter to estimate a water-leakage score representing a state of the water-leakage in the specific area.

公开(公告)号：US20170076211A1

公开(公告)日：2017-03-16

申请号：US15122461

申请日：2015-03-03

Applicant: NEC Corporation

Inventor： Yukitaka KUSUMURA ,  Ryohei FUJIMAKI ,  Yasuhiro SOGAWA ,  Satoshi MORINAGA

IPC: G06N5/04 ,  G06N99/00

CPC classification number: G06N5/04 ,  G06N7/005 ,  G06N20/00

Abstract: A feature-converting device that provides good features quickly. The device includes first and second feature construction units and first and second feature selection units. The first feature construction unit receives one or more first features and constructs one or more second features that represent the results of applying a unary function to the respective first features. The first feature selection unit computes relevance between the first and second features and a target variable that includes elements associated with elements included in the first features and selects one or more third features that represent highly relevant features. The second feature construction unit constructs one or more fourth features that represent the results of applying a multi-operand function to the third features. The second feature selection unit computes the relevance between the third and fourth features and the target variable and selects at least one fifth feature that represents highly relevant features.

Abstract translation: 功能转换设备快速提供良好的功能。 该装置包括第一和第二特征构造单元以及第一和第二特征选择单元。 第一特征构造单元接收一个或多个第一特征并构造表示将一元函数应用于相应的第一特征的结果的一个或多个第二特征。 第一特征选择单元计算第一和第二特征之间的相关性以及包括与包括在第一特征中的元素相关联的元素的目标变量，并且选择表示高度相关特征的一个或多个第三特征。 第二特征构造单元构造表示将多操作数函数应用于第三特征的结果的一个或多个第四特征。 第二特征选择单元计算第三和第四特征与目标变量之间的相关性，并且选择表示高度相关特征的至少一个第五特征。

公开(公告)号：US20170206560A1

公开(公告)日：2017-07-20

申请号：US15326273

申请日：2015-06-26

Applicant: NEC Corporation

Inventor： Yukitaka KUSUMURA ,  Hironori MIZUGUCHI ,  Ryohei FUJIMAKI ,  Satoshi MORINAGA

IPC: G06Q30/02

CPC classification number: G06Q30/0264 ,  G06Q30/02 ,  G06Q30/0202 ,  G06Q30/0242

Abstract: A prediction data input unit 91 inputs prediction data that is one or more explanatory variables that are information likely to affect future sales. An exposure pattern generation unit 92 generates an exposure pattern which is an explanatory variable indicating the content of a commercial message scheduled to be performed during a period from predicted time to future prediction target time. A component determination unit 93 determines the component used for predicting the sales, on the basis of a hierarchical latent structure that is a structure in which latent variables are represented by a tree structure and components representing probability models are located at nodes of a lowest level of the tree structure, gating functions for determining a branch direction in the nodes of the hierarchical latent structure, and the prediction data and the exposure pattern. A sales prediction unit 94 predicts the sales on the basis of the component determined by the component determination unit 93 and of the prediction data and the exposure pattern.