Google Inc.

Hugh Brendan McMahan

Daniel Ramage

Blaise Hilary Aguera-Arcas

Jakub Konecny

Xinnan Yu

Sanjiv Kumar

Li Zhang

Aaron Michael Donsbach

Krzysztof Duleba

Jon Gabriel Clapper

Juston Payne

Alison Lentz

Ananda Theertha Suresh

Jess Holbrook

Benjamin Vanik

Joshua Denali Lovejoy

Emily Anne Fortuna

Oliver Fritz Lange

Iwona Bialynicka-Birula

James Brooks Miller

Tal Dayan

Jay Pierre Civelli

Kunal Talwar

Zheng Sun

Maya Ben Ari

Christiaan Prins

Brandon Keely

Ido Ofir

Divya Tyamagundlu

Eider Brantly Moore

Tanton Holt Gibbs

Dave Morris Bacon

Ning Zheng

Rongjing Xiang

Kristine Kohlhepp

Manzil Zaheer

Sashank Jakkam Reddi

Jennifer Ellen Fernquist

Robert Douglas Fritz, Iii

John Keith Rush

Zachary Alan Garrett

Jennifer Seth

Eider Brantley Moore

Zachary Burr Charles

A computing system and method can be used to implement a version of federated learning (FL) that incorporates adaptivity (e.g., leverages an adaptive learning rate). In particular, the present disclosure provides a general optimization framework in which (1) clients perform multiple epochs of training using a client optimizer to minimize loss on their local data and (2) a server system updates its global model by applying a gradient-based server optimizer to the average of the clients' model updates. This framework can seamlessly incorporate adaptivity by using adaptive optimizers as client and/or server optimizers. Building upon this general framework, the present disclosure also provides example specific adaptive optimization techniques for FL which use per-coordinate methods as server optimizers. By focusing on adaptive server optimization, the use of adaptive learning rates is enabled without increase in client storage or communication costs and compatibility with cross-device FL can be ensured.

Federated learning with adaptive optimization

The present disclosure provides systems and methods for communication efficient distributed mean estimation. In particular, aspects of the present disclosure can be implemented by a system in which a number of vectors reside on a number of different clients, and a centralized server device seeks to estimate the mean of such vectors. According to one aspect of the present disclosure, a client computing device can rotate a vector by a random rotation matrix and then subsequently perform probabilistic quantization on the rotated vector. According to another aspect of the present disclosure, subsequent to quantization but prior to transmission, the client computing can encode the quantized vector according to a variable length coding scheme (e.g., by computing variable length codes).

Systems and methods for communication efficient distributed mean estimation

The present disclosure provides efficient communication techniques for transmission of model updates within a machine learning framework, such as, for example, a federated learning framework in which a high-quality centralized model is trained on training data distributed overt a large number of clients each with unreliable network connections and low computational power. In an example federated learning setting, in each of a plurality of rounds, each client independently updates the model based on its local data and communicates the updated model back to the server, where all the client-side updates are used to update a global model. The present disclosure provides systems and methods that reduce communication costs. In particular, the present disclosure provides at least: structured update approaches in which the model update is restricted to be small and sketched update approaches in which the model update is compressed before sending to the server.

Communication efficient federated learning

Systems and methods for learning differentially private machine-learned models are provided. A computing system can include one or more server computing devices comprising one or more processors and one or more non-transitory computer-readable media that collectively store instructions that, when executed by the one or more processors cause the one or more server computing devices to perform operations. The operations can include selecting a subset of client computing devices from a pool of available client computing devices; providing a machine-learned model to the selected client computing devices; receiving, from each selected client computing device, a local update for the machine-learned model; determining a differentially private aggregate of the local updates; and determining an updated machine-learned model based at least in part on the data-weighted average of the local updates.

Training user-level differentially private machine-learned models

Systems and methods of determining a global model are provided. In particular, one or more local updates can be received from a plurality of user devices. Each local update can be determined by the respective user device based at least in part on one or more data examples stored on the user device. The one or more data examples stored on the plurality of user devices are distributed on an uneven basis, such that no user device includes a representative sample of the overall distribution of data examples. The local updates can then be aggregated to determine a global model.

Systems and methods of distributed optimization

The present disclosure provides an image capture, curation, and editing system that includes a resource-efficient mobile image capture device that continuously captures images. The mobile image capture device is operable to input an image into at least one neural network and to receive at least one descriptor of the desirability of a scene depicted by the image as an output of the at least one neural network. The mobile image capture device is operable to determine, based at least in part on the at least one descriptor of the desirability of the scene of the image, whether to store a second copy of such image and/or one or more contemporaneously captured images in a non-volatile memory of the mobile image capture device or to discard a first copy of such image from a temporary image buffer without storing the second copy of such image in the non-volatile memory.

Systems and methods for selective retention and editing of images captured by mobile image capture device

Systems and techniques are provided for aggregation of asynchronous trust outcomes in a mobile device. Trust levels may be determined from the signals. Each trust level may be determined independently of any other trust level. Each trust level may be determined based on applying to the signals heuristics, mathematical optimization, decisions trees, machine learning systems, or artificial intelligence systems. An aggregated trust outcome may be determined by aggregating the trust levels. Aggregating the trust levels may include applying heuristics, mathematical optimization, decisions trees, machine learning systems, or artificial intelligence systems to the trust levels, and wherein the aggregated trust outcome; and sending the aggregated trust outcome to be implemented by the enabling, disabling, or relaxing of at least one security measure based on the aggregated trust outcome.

Aggregation of asynchronous trust outcomes in a mobile device

Growing community of inventors

Seattle, WA, United States of America

Hugh Brendan McMahan