公开(公告)号：US20180068016A1

公开(公告)日：2018-03-08

申请号：US15259520

申请日：2016-09-08

Applicant: XEROX CORPORATION

Inventor： Shourya Roy ,  Deepali Semwal ,  Raghuram Krishnapuram

IPC: G06F17/30

CPC classification number: G06F16/335 ,  G06F16/3329 ,  G06F16/34 ,  G09B7/02

Abstract: A method and a system for response evaluation of users from electronic documents are disclosed. In an embodiment, one or more questions and a first response pertaining to each of the one or more questions are extracted from one or more first electronic documents. Further, a second response pertaining to each of the one or more extracted questions and metadata are extracted from one or more second electronic documents. For the second response pertaining to each of the one or more extracted questions, a score is determined based on one or more similarity measures that correspond to a category of each of the one or more extracted questions. Thereafter, the response evaluation is rendered on a user interface displayed on a display screen. The response evaluation comprises at least the determined score for the second response pertaining to each of the one or more extracted questions.

公开(公告)号：US20160253597A1

公开(公告)日：2016-09-01

申请号：US14633550

申请日：2015-02-27

Applicant: Xerox Corporation

Inventor： Himanshu Sharad Bhatt ,  Deepali Semwal ,  Shourya Roy

IPC: G06N99/00

CPC classification number: G06N20/00

Abstract: An adaptation method includes using a first classifier trained on projected representations of labeled objects from a first domain to predict pseudo-labels for unlabeled objects in a second domain, based on their projected representations. A classifier ensemble is iteratively learned. The ensemble includes a weighted combination of the first classifier and a second classifier. This includes training the second classifier on the original representations of the unlabeled objects for which a confidence for respective pseudo-labels exceeds a threshold. A classifier ensemble is constructed as a weighted combination of the first classifier and the second classifier. Pseudo-labels are predicted for the remaining original representations of the unlabeled objects with the classifier ensemble and weights of the first and second classifiers in the classifier ensemble are adjusted. As the iterations proceed, the unlabeled objects progressively receive pseudo-labels which can be used for retraining the second classifier.

Abstract translation: 适应方法包括使用对来自第一域的标记对象的投影表示进行训练的第一分类器，以基于它们的投影表示来预测第二域中未标记对象的伪标签。 分类器集合被迭代地学习。 集合包括第一分类器和第二分类器的加权组合。 这包括对第二分类器对未标记对象的原始表示进行训练，对于相应伪标签的置信度超过阈值。 分类器集合被构造为第一分类器和第二分类器的加权组合。 对于具有分类器集合的未标记对象的剩余原始表示预测伪标签，并且对分类器集合中的第一和第二分类器的权重进行调整。 随着迭代的进行，未标记的对象逐渐接收可用于重新训练第二分类器的伪标签。