Amazon Technologies, Inc.

Microsoft Technology Licensing, LLC

Arun Babu Nagarajan

Andrea Olgiati

Esaias Engelbertus Greeff

Justin Allen Hutchings

Samuel David Adams

David Walker Duhon

Lakshmi Naarayanan Ramakrishnan

Erik O Peterson

Robert Franz Hain

Niket Ashok Sanghvi

J Adrian Lannin

Weixun Wang

Urvashi Chowdhary

Kyle P Wojtaszek

Stefano Stefani

Saurabh Gupta

Maximiliano Maccanti

Ramyanshu Datta

Zohar Karnin

Vladimir Zhukov

Gowda Dayananda Anjaneyapura Range

Laurence Louis Eric Rouesnel

Wei You

Owen Thomas

Kohen Berith Chia

Kunal Mehrotra

Arvind Sowmyan

Ishaaq Chandy

Uday Kumar Bandaru

Kenneth O Henderson

Zhihan Li

Khushboo Srivastava

Zhankui Lu

Dipankar Patro

Manoj Ravi

Brian Ellison Granger

Justin A Hutchings

David W Duhon

Robert F Hain

Esaias E Greeff

Niket A Sanghvi

Techniques described herein may be implemented in the context of a computing resource service provider. A machine learning (ML) service provides an interface to clients which can be used to create, read, update, and delete ML pipelines. ML pipelines can be converted to a human-readable format, which a server persists. Clients of a machine learning service can start, stop, and resume executions a ML pipeline based on the human-readable format of the ML pipeline.

Machine learning pipelines

Techniques for remote computing are described. In some examples, a decorator call in source code is used to generate a remote job to be processed by a provider network using a runtime environment that replicates a runtime environment where the source code was developed. The runtime environment is replicated using information such as runtime dependencies, packages, and environment variables of the development environment.

Local runtime replication for machine learning model training on a provider network

Parameters of a pool of computing resources to be utilized for machine learning tasks from a set of entities are stored, including a category of the computing resources, and a post-task-completion retention period during which, after completion of a task, at least a portion of data stored at the resource is not to be deleted. A compute instance of the pool is assigned to a task requested from the set of entities after determining that one or more configuration settings of the instance satisfy a preference indicated in the request for the task, and that the retention period of the instance relative to a completion of an earlier task on the instance has not expired. A result of the task is stored.

Resource management techniques to reduce startup overhead for machine learning tasks

Methods, systems, and computer-readable media for automated problem detection for machine learning models are disclosed. A machine learning analysis system receives data associated with use of a machine learning model. The data was collected by a machine learning inference system and comprises input to the model or a plurality of inferences representing output of the machine learning model. The machine learning analysis system performs analysis of the data associated with the use of the machine learning model. The machine learning analysis system detects one or more problems associated with the use of the machine learning model based at least in part on the analysis. The machine learning analysis system initiates one or more remedial actions associated with the one or more problems associated with the use of the machine learning model.

Automated problem detection for machine learning models

Device association data for peripheral devices, which is data used to connect to the peripheral devices, is maintained and synchronized across multiple computing devices associated with a user account on a service. When a new peripheral device is associated with a computing device, device association data for the new peripheral device is added to the computing device. The device association data for the new peripheral device is also sent to the service, which provides the device association to the other computing devices from which the user logs into his or her account, and those other computing devices in turn automatically associate themselves with the peripheral device. Device use data describing how a user desires to use the peripheral device on one computing device, as well as changes to the device association data on one computing device, can similarly be synchronized across the multiple computing devices.

Synchronizing device association data among computing devices

Growing community of inventors

Redmond, WA, United States of America

Arun Babu Nagarajan