公开(公告)号：US20220100721A1

公开(公告)日：2022-03-31

申请号：US17549395

申请日：2021-12-13

Applicant: Amazon Technologies, Inc.

Inventor： Nina Mishra ,  Daniel Blick ,  Sudipto Guha ,  Okke Joost Schrijvers

IPC: G06F16/215 ,  G06N5/00 ,  G06N20/00

Abstract: Random cut trees are generated with respective to respective samples of a baseline set of data records of a data set for which outlier detection is to be performed. To construct a particular random cut tree, an iterative splitting technique is used, in which the attribute along which a given set of data records is split is selected based on its value range. With respect to a newly-received data record of the stream, an outlier score is determined based at least partly on a potential insertion location of a node representing the data record in a particular random cut tree, without necessarily modifying the random cut tree.

公开(公告)号：US11003717B1

公开(公告)日：2021-05-11

申请号：US15892258

申请日：2018-02-08

Applicant: Amazon Technologies, Inc.

Inventor： Dhivya Eswaran ,  Sudipto Guha ,  Nina Mishra

IPC: G06F16/00 ,  G06F16/901 ,  G06F16/21 ,  H04L29/08 ,  H04L29/12

Abstract: Techniques for detecting anomalies in streaming graph data are described. For example, an embedding technique of generating a multi-dimensional vector of summations of each weighted edge found in both a random source bounding proper subset and a random destination bounding proper subset associated with a dimension of the epoch graph is detailed. Anomaly detection is performed on the generated multi-dimensional vectors.

公开(公告)号：US10902062B1

公开(公告)日：2021-01-26

申请号：US15686086

申请日：2017-08-24

Applicant: Amazon Technologies, Inc.

Inventor： Sudipto Guha ,  Nina Mishra

IPC: G06F16/901 ,  G06N7/00 ,  H04L12/24 ,  H04L29/06 ,  H04L12/26

Abstract: At an artificial intelligence system, a random cut tree corresponding to a sample of a multi-dimensional data set is traversed to determine a tree-specific vector indicating respective contributions of individual dimensions to an anomaly score of a particular data point. Level-specific vectors of per-dimension contributions obtained using bounding-box analyses at each level during the traversal are aggregated to obtain the tree-specific vector. An overall anomaly score contribution for at least one dimension is obtained using respective tree-specific vectors generated from one or more random cut trees, and an indication of the overall anomaly score contribution is provided.

公开(公告)号：US12174807B2

公开(公告)日：2024-12-24

申请号：US17549395

申请日：2021-12-13

Applicant: Amazon Technologies, Inc.

Inventor： Nina Mishra ,  Daniel Blick ,  Sudipto Guha ,  Okke Joost Schrijvers

IPC: G06F16/00 ,  G06F16/215 ,  G06N5/01 ,  G06N20/00 ,  G06F16/2458

Abstract: Random cut trees are generated with respective to respective samples of a baseline set of data records of a data set for which outlier detection is to be performed. To construct a particular random cut tree, an iterative splitting technique is used, in which the attribute along which a given set of data records is split is selected based on its value range. With respect to a newly-received data record of the stream, an outlier score is determined based at least partly on a potential insertion location of a node representing the data record in a particular random cut tree, without necessarily modifying the random cut tree.

公开(公告)号：US11308407B1

公开(公告)日：2022-04-19

申请号：US15842291

申请日：2017-12-14

Applicant: Amazon Technologies, Inc.

Inventor： Sudipto Guha ,  Tal Wagner ,  Shiva Prasad Kasiviswanathan ,  Nina Mishra

IPC: G06N20/00 ,  G06N5/04 ,  G06F16/901

Abstract: Examples of techniques for anomaly detection with feedback are described. An instance includes a technique is receiving a plurality of unlabeled data points from an input stream; performing anomaly detection on a point of the unlabeled data points using an anomaly detection engine; pre-processing the unlabeled data point that was subjected to anomaly detection; classifying the pre-processed unlabeled data point; determining the anomaly detection was not proper based on a comparison of a result of the anomaly detection and a result of the classifying of the pre-processed unlabeled data point; and in response to determining the anomaly detection was not proper, providing feedback to the anomaly detection engine to change at least one emphasis used in anomaly detection.

公开(公告)号：US11232085B2

公开(公告)日：2022-01-25

申请号：US14990175

申请日：2016-01-07

Applicant: Amazon Technologies, Inc.

Inventor： Nina Mishra ,  Daniel Blick ,  Sudipto Guha ,  Okke Joost Schrijvers

IPC: G06F16/215 ,  G06N5/00 ,  G06N20/00 ,  G06F16/2458

Abstract: Random cut trees are generated with respective to respective samples of a baseline set of data records of a data set for which outlier detection is to be performed. To construct a particular random cut tree, an iterative splitting technique is used, in which the attribute along which a given set of data records is split is selected based on its value range. With respect to a newly-received data record of the stream, an outlier score is determined based at least partly on a potential insertion location of a node representing the data record in a particular random cut tree, without necessarily modifying the random cut tree.

公开(公告)号：US10972491B1

公开(公告)日：2021-04-06

申请号：US15977670

申请日：2018-05-11

Applicant: Amazon Technologies, Inc.

Inventor： Sudipto Guha ,  Santosh Kalki ,  Akshay Satish

IPC: H04L29/06 ,  H04L12/26 ,  G06F9/455 ,  H04L12/46

Abstract: Techniques for seasonality-based anomaly detection and forecast are described. For example, a method of receiving a request to generate forecast for received time series data; performing a seasonality-based anomaly detection and forecast for the received time series data based upon the received request, the seasonality-based anomaly detection and forecasting to utilize a second data structure that reflect anomalies found in a first data structure on the input from the received time series data; and providing a result of the performed seasonality-based anomaly detection and forecast is described.