-
公开(公告)号：US12293262B1
公开(公告)日：2025-05-06
申请号：US16585266
申请日：2019-09-27
Applicant: Amazon Technologies, Inc.
Inventor： Andrea Olgiati , Hyunsu Cho , Rahul Iyer , Laurence Louis Eric Rouesnel
Abstract: Techniques for adaptive machine learning training via in-flight feature modification are described. A training monitor captures training data during the training of a machine learning model, and a metric generator creates metrics such as feature importance metrics based on the data. A rule evaluation engine determines whether any modification conditions are met for any of the features based on the metrics, and based on such a determination can cause the in-flight training job to be modified.
-
公开(公告)号：US11468365B2
公开(公告)日：2022-10-11
申请号：US16588930
申请日：2019-09-30
Applicant: Amazon Technologies, Inc.
Inventor： Andrea Olgiati , Rahul Raghavendra Huilgol , Vikas Kumar
Abstract: Methods, systems, and computer-readable media for GPU code injection to summarize machine learning training data are disclosed. Training of a machine learning model is initiated using a graphics processing unit (GPU) associated with a machine learning training cluster. The training of the machine learning model generates tensor data in a memory of the GPU. The GPU determines a summary of the tensor data according to a reduction operator. The summary is smaller in size than the tensor data and is output by the GPU. A machine learning analysis system performs an analysis of the training of the machine learning model based at least in part on the summary of the tensor data. The machine learning analysis system detects one or more conditions associated with the training of the machine learning model based at least in part on the analysis.
-
公开(公告)号：US11467835B1
公开(公告)日：2022-10-11
申请号：US16199129
申请日：2018-11-23
Applicant: Amazon Technologies, Inc.
Inventor： Sudipta Sengupta , Poorna Chand Srinivas Perumalla , Jalaja Kurubarahalli , Samuel Oshin , Cory Pruce , Jun Wu , Eftiquar Shaikh , Pragya Agarwal , David Thomas , Karan Kothari , Daniel Evans , Umang Wadhwa , Mark Klunder , Rahul Sharma , Zdravko Pantic , Dominic Rajeev Divakaruni , Andrea Olgiati , Leo Dirac , Nafea Bshara , Bratin Saha , Matthew Wood , Swaminathan Sivasubramanian , Rajankumar Singh
Abstract: Techniques for partitioning data flow operations between execution on a compute instance and an attached accelerator instance are described. A set of operations supported by the accelerator is obtained. A set of operations associated with the data flow is obtained. An operation in the set of operations associated with the data flow is identified based on the set of operations supported by the accelerator. The accelerator executes the first operation.
-
公开(公告)号：US10997395B2
公开(公告)日：2021-05-04
申请号：US15676015
申请日：2017-08-14
Applicant: Amazon Technologies, Inc.
Inventor： Andrea Olgiati
Abstract: Multimedia content may be obtained and an object may be identified in a first frame of video content. The object may be tracked through a plurality of frames, and the object may be identified in a second frame of the video content only if the object is no longer substantially identifiable.
-
公开(公告)号：US10534965B2
公开(公告)日：2020-01-14
申请号：US15926745
申请日：2018-03-20
Applicant: Amazon Technologies, Inc.
Inventor： Nitin Singhal , Vivek Bhadauria , Ranju Das , Gaurav D. Ghare , Roman Goldenberg , Stephen Gould , Kuang Han , Jonathan Andrew Hedley , Gowtham Jeyabalan , Vasant Manohar , Andrea Olgiati , Stefano Stefani , Joseph Patrick Tighe , Praveen Kumar Udayakumar , Renjun Zheng
Abstract: Techniques for analyzing stored video upon a request are described. For example, a method of receiving a first application programming interface (API) request to analyze a stored video, the API request to include a location of the stored video and at least one analysis action to perform on the stored video; accessing the location of the stored video to retrieve the stored video; segmenting the accessed video into chunks; processing each chunk with a chunk processor to perform the at least one analysis action, each chunk processor to utilize at least one machine learning model in performing the at least one analysis action; joining the results of the processing of each chunk to generate a final result; storing the final result; and providing the final result to a requestor in response to a second API request is described.
-
Patent Agency Ranking
公开(公告)号：US10387053B1
公开(公告)日：2019-08-20
申请号：US14856158
申请日：2015-09-16
Applicant: Amazon Technologies, Inc.
Inventor： Andrea Olgiati
IPC: G06F3/06
Abstract: Regions of memory in a distributed computing system may be synchronized. A first computing node may comprise a processor writing to a memory via a memory controller. A request to write data to the memory may be received by the memory controller. The memory controller may send a signal to a logic device which forwards the signal to other computing nodes in the distributed system. The memory controller may detect and respond to conflicting writes by instructing the computing nodes to overwrite conflicting memory regions with a data pattern indicative of the conflict.
-
公开(公告)号：US10318346B1
公开(公告)日：2019-06-11
申请号：US15274813
申请日：2016-09-23
Applicant: Amazon Technologies, Inc.
Inventor： Stavros Harizopoulos , Michail Petropoulos , Andrea Olgiati
Abstract: Data stores may implement prioritized scheduling of data store access requests. When new access requests are received, the new access requests may be scheduled for prioritized execution on processing resources. Access requests that are currently being executed with prioritized execution may be reprioritized to make additional capacity for prioritized execution of the new access requests. Prioritized execution may be automatically enabled or disabled for a data store based on monitoring of performance metrics for executing access requests.
-
公开(公告)号：US12039415B2
公开(公告)日：2024-07-16
申请号：US16588913
申请日：2019-09-30
Applicant: Amazon Technologies, Inc.
Inventor： Andrea Olgiati , Lakshmi Naarayanan Ramakrishnan , Jeffrey John Geevarghese , Denis Davydenko , Vikas Kumar , Rahul Raghavendra Huilgol , Amol Ashok Lele , Stefano Stefani , Vladimir Zhukov
Abstract: Methods, systems, and computer-readable media for debugging and profiling of machine learning model training are disclosed. A machine learning analysis system receives data associated with training of a machine learning model. The data was collected by a machine learning training cluster. The machine learning analysis system performs analysis of the data associated with the training of the machine learning model. The machine learning analysis system detects one or more conditions associated with the training of the machine learning model based at least in part on the analysis. The machine learning analysis system generates one or more alarms describing the one or more conditions associated with the training of the machine learning model.
-
公开(公告)号：US11995476B1
公开(公告)日：2024-05-28
申请号：US17482276
申请日：2021-09-22
Applicant: Amazon Technologies, Inc.
Inventor： Ramyanshu Datta , Ishaaq Chandy , Arvind Sowmyan , Wei You , Kunal Mehrotra , Kohen Berith Chia , Andrea Olgiati , Lakshmi Naarayanan Ramakrishnan , Saurabh Gupta
CPC classification number: G06F9/5038 , G06F9/5022 , G06F9/5055
Abstract: A post-task-completion retention period for which a computing resource is to be retained, without de-activating the resource, on behalf of a set of requesters of machine learning tasks is determined at a machine learning service. A first task, identified at the service prior to expiration of the retention period at a first computing resource at which a second task has completed, is initiated at the first computing resource. In response to obtaining an indication of a third task and determining that a threshold criterion associated with the retention period satisfies a criterion, the third task is initiated at an additional computing resource. The additional computing resource is de-activated after the third task completes, without waiting for the retention period to expire.
-
公开(公告)号：US11048919B1
公开(公告)日：2021-06-29
申请号：US15993222
申请日：2018-05-30
Applicant: Amazon Technologies, Inc.
Inventor： Davide Modolo , Hao Chen , Enrica Maria Filippi , Stephen Gould , Camille Claire Le Men , Andrea Olgiati
Abstract: People can be tracked across multiple segments of video data, which can correspond to different scenes in a single video file, or multiple video streams or feeds. An instance of video data can be broken up into segments that can each be analyzed to determine faces and bodies represented therein. The bodies can be analyzed across frames of the segment to determine body tracklets that are consistent across the segment. Associations of faces and bodies can be determined based using relative distances and/or spatial relationships. A subsequent clustering of these associations is performed to attempt to determine consistent associations that correspond to unique individuals. Unique identifiers are determined for each person represented in one or more segments of an instance of video data. Such an approach enables individual representations to be correlated across multiple instances.
-
-
-
-
-
-
-
-
-
Search Results
42
records
(took 0.026 seconds)
