公开(公告)号：WO2015065290A1

公开(公告)日：2015-05-07

申请号：PCT/SG2014/000508

申请日：2014-10-30

Applicant: NATIONAL UNIVERSITY OF SINGAPORE

Inventor： CHUA, Tat-Seng ,  AMIRIEBRAHIMABADI, Hadi ,  CHEN, Yan ,  CUI, Anqi

IPC: G06Q30/00

CPC classification number: G06Q10/10 ,  G06Q50/01

Abstract: Monitoring of social media and microblogs for information relevant to company brands and associated information by the use of web crawlers. A unified framework of fixed and dynamic keywords, known accounts, key users and friend lists are used to identify relevant microposts, tweets, on a particular organisation of interest. This organisational information is classified by relevancy, using learning algorithms, and emerging and evolving topics are identified.

Abstract translation: 通过使用网络抓取工具，监控社会媒体和微博，了解与公司品牌及相关信息相关的信息。 使用固定和动态关键字，已知帐户，主要用户和朋友列表的统一框架来识别感兴趣的特定组织上的相关微博，推特。 该组织信息通过相关性分类，使用学习算法，并且识别出新兴和不断发展的主题。

公开(公告)号：WO2021177897A1

公开(公告)日：2021-09-10

申请号：PCT/SG2021/050104

申请日：2021-03-03

Applicant: NATIONAL UNIVERSITY OF SINGAPORE

Inventor： FENG, Fuli ,  RUI, Xilin ,  WANG, Wenjie ,  CAO, Yixin ,  CHUA, Tat-Seng

IPC: G06N3/08 ,  G06F40/279

Abstract: Described herein is a method for training a neural network. The method includes pre-training the neural network based on at least one of: (i) a magnitude model, whereby pre-training comprises: masking a plurality of said numbers (masked numbers); predicting a magnitude of each masked number; and minimising a total loss over the masked numbers; and (ii) a value model, whereby pre-training comprises: partially masking a plurality of said numbers (partially masked numbers) such that each partially masked number comprises a visible portion; predicting a value of each partially masked number; and minimising a total loss over the partially masked numbers. The method also includes training the pre-trained neural network on a domain-specific corpus and then applying the trained neural network to a document to identify a relationship between one or more numbers in the document and a context of the document.

公开(公告)号：WO2018212711A1

公开(公告)日：2018-11-22

申请号：PCT/SG2018/050234

申请日：2018-05-15

Applicant: NATIONAL UNIVERSITY OF SINGAPORE

Inventor： HE, Xiangnan ,  CHUA, Tat-Seng

IPC: G06N3/02 ,  G06F15/18 ,  G06N7/00

CPC classification number: G06F15/76 ,  G06N3/0454 ,  G06N3/0481 ,  G06N3/082

Abstract: Methods and systems for predictive analysis are disclosed, A predictive analysis method comprises: receiving a set of predictor variables as an input feature vector comprising a plurality of features; projecting each feature of the feature vector onto a dense vector representation to obtain a set of embedding vectors representing the input feature vector in an embedding space; converting the set of embedding vectors into a bi-interaction pooling vector that encodes second-order interactions between features of the feature vector in the embedding space; inputting the bi-interaction pooling vector into a hidden layer stack, the hidden layer stack comprising at least one hidden layer of neural network nodes; and transforming an output vector of the hidden layer stack into a prediction score.

公开(公告)号：WO2021145823A1

公开(公告)日：2021-07-22

申请号：PCT/SG2021/050022

申请日：2021-01-13

Applicant: NATIONAL UNIVERSITY OF SINGAPORE

Inventor： LEI, Wenqiang ,  HE, Xiangnan ,  MIAO, Yisong ,  KAN, Min-Yen ,  CHUA, Tat-Seng

IPC: G06F16/332 ,  G06N20/00 ,  G06Q30/00

Abstract: Disclosed is a method for refining item recommendations including: (a) receiving a session initiation from the user; (b) ranking, recommender component(s) (RC) and based on a conversation history of the user, candidate items (candidate items) and item attributes (candidate attributes) for the user; and (c) determining, using conversational component(s) (CC) and based on the ranked candidate items, ranked candidate attributes and conversation history, either to: ask a said ranked candidate attribute; or recommend at least one said ranked candidate item. In the former case, a candidate attribute is asked (the asked attribute). In the latter case, a candidate item is recommended (the recommended item). The method also includes receiving acceptance or rejection of the asked attribute or recommended item by the user, feeding the acceptance or rejection to the RC and updating the conversation history, and either exiting or repeating from step (c).

公开(公告)号：WO2019035771A1

公开(公告)日：2019-02-21

申请号：PCT/SG2018/050420

申请日：2018-08-17

Applicant: NATIONAL UNIVERSITY OF SINGAPORE

Inventor： REN, Tongwei ,  GUO, Jingfan ,  CHUA, Tat-Seng ,  SHANG, Xindi

IPC: G06K9/46 ,  G06T7/20

Abstract: Methods and systems for detecting visual relations in a video are disclosed. A method comprises: decomposing the video sequence into a plurality of segments; for each segment, detecting objects in frames of the segment; tracking the detected objects over the segment to form a set of object tracklets for the segment; for the detected objects, extracting object features; for pairs of object tracklets of the set of object tracklets, extracting relativity features indicative of a relation between the objects corresponding to the pair of object tracklets; forming relation feature vectors for pairs of object tracklets using the object features of objects corresponding to respective pairs of object tracklets and the relativity features of the respective pairs of object tracklets; and generating a set of segment relation prediction results from the relation features vectors; generating a set of visual relation instances for the video sequence by merging the segment prediction results from different segments; and generating a set of visual relation detection results from the set of visual relation instances.

公开(公告)号：WO2018212710A1

公开(公告)日：2018-11-22

申请号：PCT/SG2018/050233

申请日：2018-05-15

Applicant: NATIONAL UNIVERSITY OF SINGAPORE

Inventor： HE, Xiangnan ,  ZHANG, Hanwang ,  CHUA, Tat-Seng

IPC: G06F15/18 ,  G06N7/00 ,  G06N3/02

CPC classification number: G06F15/76 ,  G06N3/04 ,  G06N3/082

Abstract: Methods and systems for predictive analysis are disclosed, A predictive analysis method comprises: receiving a set of predictor variables as an input feature vector comprising a plurality of features; projecting each feature of the input feature vector onto a dense vector representation to obtain a set of embedding vectors presenting the input feature vector in an embedding space; calculating a set of interacted vectors, each interacted vector being an element-wise product of two embedding vectors of the set of embedding vectors; performing a weight sum over the interacted vectors, the weighted sum being weighted by a plurality of attention scores each corresponding to an interaction between a pair of features of the feature vector; and projecting the weighed sum to obtain a prediction score.