公开(公告)号：US10354199B2

公开(公告)日：2019-07-16

申请号：US14960869

申请日：2015-12-07

Applicant: Xerox Corporation

Inventor： Stéphane Clinchant ,  Gabriela Csurka ,  Boris Chidlovskii

IPC: G06N20/00 ,  G06N3/04

Abstract: A classification method includes receiving a collection of samples, each sample comprising a multidimensional feature representation. A class label prediction for each sample in the collection is generated with one or more pretrained classifiers. For at least one iteration, each multidimensional feature representation is augmented with a respective class label prediction to form an augmented representation, a set of corrupted samples is generated from the augmented representations, and a transformation that minimizes a reconstruction error for the set of corrupted samples is learned. An adapted class label prediction for at least one of the samples in the collection is generated using the learned transformation and information is output, based on the adapted class label prediction. The method is useful in predicting labels for target samples where there is no access to source domain samples that are used to train the classifier and no access to target domain training data.

公开(公告)号：US20180347999A1

公开(公告)日：2018-12-06

申请号：US15607882

申请日：2017-05-30

Applicant: Xerox Corporation

Inventor： Boris Chidlovskii

IPC: G01C21/34 ,  G06Q10/04 ,  G08G1/01 ,  G06F17/50

Abstract: Method and apparatus for generating realistic samples of public transportation usage to improve the operability of a public transportation system. Constraints can be expressed as a group of origin-destination-time triples. A trip (or trips) can then be assigned to each triple among the group of origin-destination-time triples while ignoring capacity constraints. A Metropolis-Hasting class sampling technique can then be applied with respect to the trip beginning with the origin-destination-time triples to generate a realistic sample of public transportation usage based on the aforementioned constraints in the form of target probability distributions and/or target probability densities, thereby improving the public transportation system by taking into account the generated realistic sample of public transportation usage.

公开(公告)号：US09916542B2

公开(公告)日：2018-03-13

申请号：US15013273

申请日：2016-02-02

Applicant: Xerox Corporation

Inventor： Boris Chidlovskii ,  Gabriela Csurka ,  Stéphane Clinchant

IPC: G06N99/00 ,  G06K9/62 ,  G06K9/00

CPC classification number: G06N99/005 ,  G06K9/00684 ,  G06K9/6256 ,  G06K9/6262 ,  G06K9/6298

Abstract: A machine learning method operates on training instances from a plurality of domains including one or more source domains and a target domain. Each training instance is represented by values for a set of features. Domain adaptation is performed using stacked marginalized denoising autoencoding (mSDA) operating on the training instances to generate a stack of domain adaptation transform layers. Each iteration of the domain adaptation includes corrupting the training instances in accord with feature corruption probabilities that are non-uniform over at least one of the set of features and the domains. A classifier is learned on the training instances transformed using the stack of domain adaptation transform layers. Thereafter, a label prediction is generated for an input instance from the target domain represented by values for the set of features by applying the classifier to the input instance transformed using the stack of domain adaptation transform domains.

公开(公告)号：US20170220951A1

公开(公告)日：2017-08-03

申请号：US15013401

申请日：2016-02-02

Applicant: Xerox Corporation

Inventor： Boris Chidlovskii ,  Gabriela Csurka ,  Stéphane Clinchant

IPC: G06N99/00 ,  G06F17/27

CPC classification number: G06N20/00 ,  G06F16/35

Abstract: Training instances from a target domain are represented by feature vectors storing values for a set of features, and are labeled by labels from a set of labels. Both a noise marginalizing transform and a weighting of one or more source domain classifiers are simultaneously learned by minimizing the expectation of a loss function that is dependent on the feature vectors corrupted with noise represented by a noise probability density function, the labels, and the one or more source domain classifiers operating on the feature vectors corrupted with the noise. An input instance from the target domain is labeled with a label from the set of labels by operations including applying the learned noise marginalizing transform to an input feature vector representing the input instance and applying the one or more source domain classifiers weighted by the learned weighting to the input feature vector representing the input instance.

公开(公告)号：US09519060B2

公开(公告)日：2016-12-13

申请号：US14287538

申请日：2014-05-27

Applicant: Xerox Corporation

Inventor： Boris Chidlovskii ,  Gabriela Csurka ,  Jose Antonio Rodriguez Serrano

IPC: G01C3/08 ,  G01S17/88 ,  G01S17/42 ,  G01S7/48 ,  G06T7/00 ,  G01S17/89 ,  G06K9/00

CPC classification number: G01S17/88 ,  G01S7/4802 ,  G01S17/42 ,  G01S17/89 ,  G06K9/00805 ,  G06K9/6206 ,  G06T7/00 ,  G06T7/30 ,  G06T2207/10028

Abstract: A system and method for classifying vehicles from laser scan data by receiving laser scan data corresponding to multiple vehicles from a laser scanner; extracting vehicle shapes corresponding to the multiple vehicles based on the laser scan data; aligning the vehicle shapes; and generating vehicle profiles based on the aligned vehicle shapes. The system and method can further include aligning the vehicle shapes using sequence kernels, such as global alignment kernels, and constraining the sequence kernels based on determined weights.

Abstract translation: 一种用于通过从激光扫描仪接收对应于多个车辆的激光扫描数据来对车辆进行激光扫描数据分类的系统和方法; 基于所述激光扫描数据提取与所述多个车辆相对应的车辆形状; 对准车辆形状; 以及基于对齐的车辆形状产生车辆轮廓。 系统和方法还可以包括使用诸如全局对准内核之类的序列内核来对齐车辆形状，并且基于确定的权重来约束序列内核。

公开(公告)号：US09116894B2

公开(公告)日：2015-08-25

申请号：US13828048

申请日：2013-03-14

Applicant: Xerox Corporation

Inventor： Boris Chidlovskii

IPC: G06F17/30

CPC classification number: G06F17/30038 ,  G06F17/30265

Abstract: A method and system is disclosed for tagging a latent object with selected tag recommendations, including a set of content objects wherein each object is characterized by an associated set of content features. An annotation relationship is determined between the features and a pre-determined tag for the each object, the relationship being defined by a graph construction representative of an affinity relationship between each pre-selected tag and content object to a selected query. A plurality of the annotation relationships are ranked based upon a relevance of the preselected tags to the content features in response to a new query for assigning a new tag to the each object, so that a suggested tag is made from the ranking whereby the suggested tag is determined as a most likely tag for annotating the content object.

Abstract translation: 公开了一种用于使用选定的标签推荐来标记潜在对象的方法和系统，包括一组内容对象，其中每个对象由相关联的内容特征集合表征。 在特征和每个对象的预定标签之间确定注释关系，该关系由表示每个预先选择的标签与内容对象与所选查询之间的亲和度关系的图形构造来定义。 响应于向每个对象分配新标签的新查询，基于预选标签与内容特征的相关性来对多个注释关系进行排名，从而从推荐标签中进行建议标签 被确定为用于注释内容对象的最可能的标签。

公开(公告)号：US20140280232A1

公开(公告)日：2014-09-18

申请号：US13828048

申请日：2013-03-14

Applicant: XEROX CORPORATION

Inventor： Boris Chidlovskii

IPC: G06F17/30

CPC classification number: G06F17/30038 ,  G06F17/30265

Abstract: A method and system is disclosed for tagging a latent object with selected tag recommendations, including a set of content objects wherein each object is characterized by an associated set of content features. An annotation relationship is determined between the features and a pre-determined tag for the each object, the relationship being defined by a graph construction representative of an affinity relationship between each pre-selected tag and content object to a selected query. A plurality of the annotation relationships are ranked based upon a relevance of the preselected tags to the content features in response to a new query for assigning a new tag to the each object, so that a suggested tag is made from the ranking whereby the suggested tag is determined as a most likely tag for annotating the content object.

Abstract translation: 公开了一种用于使用选定的标签推荐来标记潜在对象的方法和系统，包括一组内容对象，其中每个对象由相关联的内容特征集合表征。 在特征和每个对象的预定标签之间确定注释关系，该关系由表示每个预先选择的标签与内容对象与所选查询之间的亲和度关系的图形构造来定义。 响应于向每个对象分配新标签的新查询，基于预先选择的标签与内容特征的相关性来对多个注释关系进行排名，从而从推荐的标签中进行建议的标签， 被确定为用于注释内容对象的最可能的标签。

公开(公告)号：US10296846B2

公开(公告)日：2019-05-21

申请号：US14950544

申请日：2015-11-24

Applicant: Xerox Corporation

Inventor： Gabriela Csurka ,  Boris Chidlovskii ,  Stéphane Clinchant

IPC: G06N3/04 ,  G06N3/08 ,  G06N5/02 ,  G06N20/00

Abstract: A domain-adapted classification system and method are disclosed. The method includes mapping an input set of representations to generate an output set of representations, using a learned transformation. The input set of representations includes a set of target samples from a target domain. The input set also includes, for each of a plurality of source domains, a class representation for each of a plurality of classes. The class representations are representative of a respective set of source samples from the respective source domain labeled with a respective class. The output set of representations includes an adapted representation of each of the target samples and an adapted class representation for each of the classes for each of the source domains. A class label is predicted for at least one of the target samples based on the output set of representations and information based on the predicted class label is output.

公开(公告)号：US20170220897A1

公开(公告)日：2017-08-03

申请号：US15013273

申请日：2016-02-02

Applicant: Xerox Corporation

Inventor： Boris Chidlovskii ,  Gabriela Csurka ,  Stéphane Clinchant

IPC: G06K9/62 ,  G06N99/00

CPC classification number: G06N99/005 ,  G06K9/00684 ,  G06K9/6256 ,  G06K9/6262 ,  G06K9/6298

Abstract: A machine learning method operates on training instances from a plurality of domains including one or more source domains and a target domain. Each training instance is represented by values for a set of features. Domain adaptation is performed using stacked marginalized denoising autoencoding (mSDA) operating on the training instances to generate a stack of domain adaptation transform layers. Each iteration of the domain adaptation includes corrupting the training instances in accord with feature corruption probabilities that are non-uniform over at least one of the set of features and the domains. A classifier is learned on the training instances transformed using the stack of domain adaptation transform layers. Thereafter, a label prediction is generated for an input instance from the target domain represented by values for the set of features by applying the classifier to the input instance transformed using the stack of domain adaptation transform domains.

公开(公告)号：US20160253681A1

公开(公告)日：2016-09-01

申请号：US14632207

申请日：2015-02-26

Applicant: Xerox Corporation

Inventor： Cesar Roberto de Souza ,  Boris Chidlovskii ,  Frederic Roulland ,  Luis Rafael Ulloa Paredes

IPC: G06Q30/02 ,  G06F17/30 ,  G06N99/00

CPC classification number: G06Q30/0201 ,  G06F17/30572 ,  G06F17/30598

Abstract: Methods and systems for interpretable user behavior profiling in off-street parking applications. To render user profiles easy to interpret by decision makers, the semi-automatic discovery and tagging of user profiles can be implemented. Transaction data from one or more (and geographically close) off-street parking installations can be implemented. An analysis of spatio-temporal behavioral patterns can be implemented based on representation of any parking episode by a set of heterogeneous features, the use of clustering methods for automatic pattern discovery, an assessment of obtained clusters, semi-automatic identification/tagging of space-temporal patterns, and a user-friendly interpretation of obtained patterns.

Abstract translation: 在路边停车场应用中解释用户行为剖析的方法和系统。 为了使决策者易于理解用户简档，可以实现用户配置文件的半自动发现和标记。 可以实施来自一个或多个（和地理上接近）的路边停车设施的交易数据。 时空行为模式的分析可以基于一组异构特征表示任何停车情节，使用自动模式发现的聚类方法，获得的群集的评估，空间 - 时间行为模式的半自动识别/ 时间模式，以及对所获取模式的用户友好解释。