公开(公告)号：US11748568B1

公开(公告)日：2023-09-05

申请号：US16988153

申请日：2020-08-07

Applicant: Amazon Technologies, Inc.

Inventor： Umut Orhan ,  Harshad Vasant Kulkarni ,  Jasmeet Chhabra ,  Vikas Dharia

IPC: G06F40/289 ,  G06F40/284 ,  H04L41/16 ,  H04L43/16 ,  H04L43/024

CPC classification number: G06F40/284 ,  H04L41/16 ,  H04L43/16 ,  H04L43/024

Abstract: A plurality of metrics records, including some records indicating metrics for which anomaly analysis has been performed, is obtained. Using a training data set which includes the metrics records, a machine learning model is trained to predict an anomaly analysis relevance score for an input record which indicates a metric name. Collection of a particular metric of an application is initiated based at least in part on an anomaly analysis relevance score obtained for the particular metric using a trained version of the model.

公开(公告)号：US12073182B2

公开(公告)日：2024-08-27

申请号：US18353164

申请日：2023-07-17

Applicant: Amazon Technologies, Inc.

Inventor： Umut Orhan ,  Harshad Vasant Kulkarni ,  Jasmeet Chhabra ,  Vikas Dharia

IPC: G06F40/284 ,  H04L41/16 ,  H04L43/16 ,  H04L43/024

CPC classification number: G06F40/284 ,  H04L41/16 ,  H04L43/16 ,  H04L43/024

Abstract: A plurality of metrics records, including some records indicating metrics for which anomaly analysis has been performed, is obtained. Using a training data set which includes the metrics records, a machine learning model is trained to predict an anomaly analysis relevance score for an input record which indicates a metric name. Collection of a particular metric of an application is initiated based at least in part on an anomaly analysis relevance score obtained for the particular metric using a trained version of the model.

公开(公告)号：US20240028830A1

公开(公告)日：2024-01-25

申请号：US18353164

申请日：2023-07-17

Applicant: Amazon Technologies, Inc.

Inventor： Umut Orhan ,  Harshad Vasant Kulkarni ,  Jasmeet Chhabra ,  Vikas Dharia

IPC: G06F40/284 ,  H04L41/16 ,  H04L43/16

CPC classification number: G06F40/284 ,  H04L41/16 ,  H04L43/16 ,  H04L43/024

Abstract: A plurality of metrics records, including some records indicating metrics for which anomaly analysis has been performed, is obtained. Using a training data set which includes the metrics records, a machine learning model is trained to predict an anomaly analysis relevance score for an input record which indicates a metric name. Collection of a particular metric of an application is initiated based at least in part on an anomaly analysis relevance score obtained for the particular metric using a trained version of the model.