公开(公告)号：WO2022221488A2

公开(公告)日：2022-10-20

申请号：PCT/US2022/024750

申请日：2022-04-14

Applicant: CAPITAL ONE SERVICES, LLC

Inventor： LAHLOU, Tarek Aziz ,  DELAUNAY, Megan Lynn ,  FYOCK, Corey Jonathan ,  BABINSKY, Erin

IPC: G06N3/08 ,  G06N20/00 ,  G06N3/084 ,  G06N5/04

Abstract: Systems, apparatuses, and methods are described for data labeling for training artificial intelligence systems. A candidate dataset comprising data samples and corresponding labels may be used to update an incumbent dataset comprise data samples and corresponding labels. The integrity of a data sample-label pair in the candidate dataset may be determined before the data sample-label pair is added to the incumbent dataset. For determining labeling integrity, a plurality of machine classifiers may be trained based on the incumbent dataset and portions of the candidate dataset. The plurality of machine classifiers as trained may be used to generate predicted labels for data samples in the candidate dataset. The integrity of the data sample-label pair in the candidate dataset may be measured based on the predicted labels for the data sample.

公开(公告)号：WO2022212579A1

公开(公告)日：2022-10-06

申请号：PCT/US2022/022646

申请日：2022-03-30

Applicant: AMAZON TECHNOLOGIES, INC.

Inventor： COOK, John Byron ,  NOETZLI, Andres Philipp ,  RUNGTA, Neha ,  HU, Jingmei

IPC: G06F9/50 ,  G06N5/00 ,  G06N5/04 ,  G06F2209/5017 ,  G06F30/33 ,  G06F9/5027 ,  G06F9/5038 ,  G06F9/505 ,  G06F9/5072 ,  G06F9/541 ,  G06N5/006 ,  G06N5/043

Abstract: Techniques are described for efficiently distributing across multiple computing resources satisfiability modulo theories (SMT) queries expressed in propositional logic with string variables. As part of the computing-related services provided by a cloud provider network, many cloud providers also offer identity and access management services, which generally help users to control access and permissions to the services and resources (e.g., compute instances, storage resources, etc.) obtained by users via a cloud provider network. By using resource policies, for example, users can granularly control which identities are able to access specific resources associated with the users' accounts and how those identities can use the resources. The ability to efficiently distribute the analysis of SMT queries expressed in propositional logic with string variables among any number of separate computing resources (e.g., among separate processes, compute instances, containers, etc.) enables the efficient analysis of such policies.

公开(公告)号：WO2022209498A1

公开(公告)日：2022-10-06

申请号：PCT/JP2022/008062

申请日：2022-02-25

Applicant: ソニーグループ株式会社

Inventor： 佐塚　直也

IPC: G06Q10/06 ,  G06N3/02 ,  G06N5/04

Abstract: 本開示の一側面に係る生体情報処理装置は、導出部と、分類部とを備えている。導出部は、特定タスクを実行している最中の対象生体から得られた生体情報および行動情報の少なくとも１つに基づいて対象生体の情動情報を導出する。分類部は、所定の分類指標に基づいて、導出部で得られた情動情報を分類する。

公开(公告)号：WO2022206197A1

公开(公告)日：2022-10-06

申请号：PCT/CN2022/076329

申请日：2022-02-15

Applicant: INTERNATIONAL BUSINESS MACHINES CORPORATION ,  IBM (CHINA) CO., LIMITED

Inventor： LUO, Tong ,  DAI, Yi ,  ZHANG, Guangming ,  SUN, Bingjiang ,  CAO, Shunxin ,  CHEN, Yan ,  ZHUO, Ling

IPC: G06N20/00 ,  G06N5/04

Abstract: A method is presented to facilitate the training of a very large number of machine-learning performance models used to detect anomalies in computing operations. The models are grouped together according to model type, and are allocated to different pods of a computing environment that is used to carry out the operations being monitored. Initial training of models in a group is carried out while monitoring resource usage, and a particular pod is selected for further training based on the resource usage. The pod selected for training preferably has a minimum change in resource usage before and after the initial training. A different pod can be selected for scoring the trained models. The pod selected for scoring preferably has a maximum resource usage during an initial scoring among all pods.

公开(公告)号：WO2022204435A2

公开(公告)日：2022-09-29

申请号：PCT/US2022/021801

申请日：2022-03-24

Applicant: TRUST & SAFETY LABORATORY INC.

Inventor： SIEGEL, Thomas ,  PONNEKANTI, Shankar, R. ,  LONEY, Benjamin, P.

IPC: G06F16/30 ,  G06F16/9535 ,  G06F15/173 ,  G06F16/903 ,  G06F16/95 ,  G06F40/216 ,  G06N5/02 ,  G06N5/04

Abstract: A system and method are provided for detecting, measuring, and/or mitigating contentious multi-platform content. The method includes recording any contentious content in one or more online platforms tagged by a plurality of users, while the plurality of users are searching the one or more online platforms according to a specified criteria. The method also includes analyzing actions of the one or more online platforms to determine an extent of contentious content in the one or more online platforms tagged by the plurality of users. The method also includes generating a report indicating the extent of contentious content, for the one or more online platforms. In some implementations, the method also includes providing, to the plurality of users, an interface that specifies the criteria for identifying contentious content in the one or more online platforms.

公开(公告)号：WO2022177345A1

公开(公告)日：2022-08-25

申请号：PCT/KR2022/002418

申请日：2022-02-18

Applicant: 주식회사 인포플라

Inventor： 최인묵

IPC: G06F9/54 ,  G06F9/455 ,  G06F9/48 ,  G06N5/04 ,  G06N20/00 ,  G06F3/0489 ,  G06V10/10

Abstract: AI기반으로 화면 정보를 인지하여 화면 상의 오브젝트에 이벤트를 발생시키는 방법은 사용자 PC에서 웹기반 IT운영관리시스템 플랫폼에 접속하여 스케줄러를 등록하는 것, 스케줄러가 등록되면 웹기반 IT운영관리시스템 플랫폼의 AI웹소켓에 스케줄러의 등록을 알리는 것, 웹기반 IT운영관리시스템 플랫폼의 AI웹소켓으로부터 정해진 시간에 사용자 PC의 AI스크린 에이전트(Agent)의 AI웹소켓으로 통신을 통해 스케줄러의 시작을 알리는 데이터를 전송하는 것, AI스크린 에이전트가 웹기반 IT운영관리시스템 플랫폼의 AI스크린으로 사용자 PC의 화면 이미지를 전송하고 학습된 AI 모델을 포함하는 AI스크린으로부터 화면 상의 하나 이상의 오브젝트의 위치를 추론한 정보데이터를 요청하는 것, AI스크린이 전송받은 화면 이미지로부터 AI스크린의 학습된 AI 모델을 통해 화면의 하나 이상의 오브젝트의 위치를 추론하는 것, 추론된 하나 이상의 오브젝트의 위치에 대한 정보데이터를 AI스크린 에이전트의 AI웹소켓으로 통신을 통해 전송하는 것, 및 AI스크린 에이전트가 전송된 데이터를 토대로 사용자 pc의 화면에서 하나 이상의 오브젝트에 대한 이벤트를 발생시키는 것을 포함할 수 있다.

公开(公告)号：WO2022170358A1

公开(公告)日：2022-08-11

申请号：PCT/US2022/070547

申请日：2022-02-07

Applicant: SCHLUMBERGER TECHNOLOGY CORPORATION ,  SCHLUMBERGER CANADA LIMITED ,  SERVICES PETROLIERS SCHLUMBERGER ,  GEOQUEST SYSTEMS B.V.

Inventor： MUSTAPHA, Hussein ,  SU, Shi ,  RAMATULLAYEV, Samat ,  DIAS, Daniel

IPC: G01V1/28 ,  G06Q50/02 ,  G01V1/50 ,  G01V1/48 ,  G06F17/18 ,  G06N5/04 ,  E21B43/40 ,  G06N20/00 ,  G06K9/62

Abstract: A method includes receiving data representing reservoir properties for a subsurface volume, conducting an uncertainty analysis by simulating different model realizations representing the subsurface volume, identifying a first hot spot of the subsurface volume based on the uncertainty analysis, the first hot spot representing an area having a high predicted performance, relative to other areas of the subsurface volume, based on the simulating of the different model realizations representing the subsurface volume, identifying a second hot spot of the subsurface volume using a machine learning model that is trained to predict well performance based on the one or more reservoir properties, evaluating the first and second hot spots for well placement based on the predicted well performance at the first and second hot spots, respectively, and selecting at least one of the first hot spot or the second hot spot for well construction.

公开(公告)号：WO2022163985A1

公开(公告)日：2022-08-04

申请号：PCT/KR2021/017320

申请日：2021-11-23

Applicant: 주식회사 노타

Inventor： 김태호 ,  채명수 ,  백종원 ,  김동욱 ,  임동욱 ,  박철빈 ,  차우다리비벡

IPC: G06N5/04 ,  G06N5/02

Abstract: 인공지능 모델 경량화 방법 및 시스템을 개시한다. 일실시예에 따른 경량화 방법은, 경량화를 위한 추론모델을 입력받는 단계, 타겟 디바이스 풀에서 타겟 디바이스를 선택하는 단계, 압축 메서드 풀에서 압축 메서드의 조합을 선택하는 단계, 추론모델을 선택된 압축 메서드의 조합을 이용하여 압축하는 단계, 선택된 타겟 디바이스를 이용하여 압축된 추론모델의 성능을 측정하는 단계 및 측정된 성능에 기반하여 최종 경량화 추론모델을 결정하는 단계를 포함할 수 있다.

公开(公告)号：WO2022149247A1

公开(公告)日：2022-07-14

申请号：PCT/JP2021/000394

申请日：2021-01-07

Applicant: 日本電信電話株式会社

Inventor： 高橋　元悟

IPC: G06F11/36 ,  G06N5/04

Abstract: 希少手順生成装置は、予測部と、フィルタリング処理部と、抽出処理部とを備える。予測部は、複数の機能を有するシステムにおける機能の利用順序が記録されたログである学習用の第１のデータに基づいて機能の利用順序を学習した学習モデルを用いて、機能の利用順序のうちで一部の順序の機能が欠損しているログである予測用の第２のデータから、欠損している機能を予測確率とともに予測する。フィルタリング処理部は、予測確率がゼロよりも大きくて閾値以下であるか否かを判定する。抽出処理部は、フィルタリング処理部の判定結果に従って第２のデータを抽出する。

公开(公告)号：WO2022147518A1

公开(公告)日：2022-07-07

申请号：PCT/US2022/011071

申请日：2022-01-04

Applicant: EDGECORTIX PTE. LTD. ,  EDGECORTIX CORPORATION

Inventor： NEZ, Nikolay ,  ZOHOURI, Hamid Reza ,  KHAVIN, Oleg ,  VILCHEZ, Antonio Tomas Nevado ,  DASGUPTA, Sakyasingha

IPC: G06N3/063 ,  G06F17/15 ,  G06N5/04

Abstract: Neural network inference may be performed by configuration of a device including an accumulation memory, a plurality of convolution modules configured to perform mathematical operations on input values, a plurality of adder modules configured to sum values output from the plurality of convolution modules, and a plurality of convolution output interconnects connecting the plurality of convolution modules, the plurality of adder modules, and the accumulation memory. The accumulation memory is an accumulation memory allocation of a writable memory block having a reconfigurable bank width, and each bank of the accumulation memory allocation is a virtual combination of consecutive banks of the writable memory block.