公开(公告)号：WO2022123380A1

公开(公告)日：2022-06-16

申请号：PCT/IB2021/060921

申请日：2021-11-24

Applicant: INTERNATIONAL BUSINESS MACHINES CORPORATION [US]/[US] ,  IBM (CHINA) INVESTMENT COMPANY LTD. [CN]/[CN] (MG) ,  IBM ISRAEL - SCIENCE & TECHNOLOGY LTD. [IL]/[IL] (MG)

Inventor： KATSUKI, Takayuki ,  OSOGAMI, Takayuki

IPC: G06N3/04

Abstract: A method, a computer program product, and a system of adversarial semi-supervised one-shot training using a data stream. The method includes receiving a data stream based on an observation, wherein the data stream includes unlabeled data and labeled data. The method also includes training a prediction model with the labeled data using stochastic gradient descent based on a classification loss and an adversarial term and training a representation model with the labeled data and the unlabeled data based on a reconstruction loss and the adversarial term. The adversarial term is a cross-entropy between the middle layer output data from the models. The classification loss is a cross-entropy between the labeled data and an output from the prediction model. The method further includes updating a discriminator with middle layer output data from the prediction model and the representation model and based on a discrimination loss, and discarding the data stream.

公开(公告)号：WO2022123370A1

公开(公告)日：2022-06-16

申请号：PCT/IB2021/060784

申请日：2021-11-21

Applicant: INTERNATIONAL BUSINESS MACHINES CORPORATION [US]/[US] ,  IBM ISRAEL SCIENCE AND TECHNOLOGY LTD. [IL]/[IL] (MG) ,  IBM (CHINA) INVESTMENT COMPANY LTD. [CN]/[CN] (MG)

Inventor： ROONEY, John ,  GARCÉS ERICE, Luis

IPC: G06F16/21

Abstract: An approach to finding data locations may be provided. A first synopses, which corresponds to first tabular data (14) may be provided. An auxiliary data storage system (2) may be scanned to identify second tabular data (24) stored in the auxiliay data storage system (2). A second synopses may be obtained, in which the synopses correspond to the second columns of second tabular data (24). The synopsis may be computed for each second column of the second columns, according to a numeric representation of contents of cells of said each second column. The computed synopsis may include a vector of m descriptors. The two sets of one or more descriptors may be compared. A subset of the descriptors of the second synopses may be compared with corresponding descriptors of the first synopses, to identify potential matches between the second tabular data (24) and the corpus of first tabular data (14).

公开(公告)号：WO2022123437A1

公开(公告)日：2022-06-16

申请号：PCT/IB2021/061407

申请日：2021-12-07

Applicant: INTERNATIONAL BUSINESS MACHINES CORPORATION [US]/[US] ,  IBM (CHINA) INVESTMENT COMPANY LTD. [CN]/[CN] (MG) ,  IBM DEUTSCHLAND GMBH [DE]/[DE] (MG)

Inventor： BÖHME, Thomas ,  SCHILLER, Oliver ,  FINNIE, James ,  SCHUSTER, Ingo

IPC: G06F16/90

Abstract: A method, a computer program product, and a system for scheduling a first query execution plan for executing a first query for accessing first data sets of data of a relational database system (RDBS) and a second query execution plan for executing a second query for accessing second data sets of the data. The data is divided into data blocks of data sets of the data, wherein metadata of the RDBS comprises respective information about at least one property of the data sets of a respective data block of the data blocks. The scheduling may be performed on the basis of an examination if there is a common partition between a first partition of data blocks which may fulfill a first condition according to the first query and a second partition of data blocks which may fulfill a second condition according to the second query.