Google Inc.

New York University

Oracle International Corporation

Chen Wang

David G Grier

Erik Arjan Hendriks

Debargha Mukherjee

Michele Mae Covell

Hui Xu

Dilip Krishnan

Sarah Parker

Fook Chiong Cheong

Ramin D Zabih

Michael D Ward

David B Ruffner

Aaron Joseph Maschinot

Laura A Philips

Aaron Yehuda Sarna

Piotr Teterwak

Xiao Zhong

Yonglong Tian

Emil Keyder

Prannay Khosla

Ce Liu

Jonathan Chew

Nicolas Michael

Andrew Hollingsworth

Thomas William Knych

Hagay Shpaisman

Phillip John Isola

Mario Guajardo-Cespedes

Philip John Isola

Erik Hendriks

The present disclosure provides an improved training methodology that enables supervised contrastive learning to be simultaneously performed across multiple positive and negative training examples. In particular, example aspects of the present disclosure are directed to an improved, supervised version of the batch contrastive loss, which has been shown to be very effective at learning powerful representations in the self-supervised setting. Thus, the proposed techniques adapt contrastive learning to the fully supervised setting and also enable learning to occur simultaneously across multiple positive examples.

Supervised contrastive learning with multiple positive examples

Systems and methods for holographic characterization of protein aggregates. Size and refractive index of individual aggregates in a solution can be determined. Information regarding morphology and porosity can be extracted from holographic data.

Holographic characterization of protein aggregates

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for verifying consistency between content of a native application and content of a corresponding resource that is provided separately from the content of the native application.

Native application content verification

An apparatus for encoding a current frame of a video. The apparatus includes a memory and a processor. The processor is configured to execute instructions stored in the memory to generate, for each reference frame of a subset of available reference frames, at least one respective candidate global motion model (GMM); partition the current frame into blocks; generate an aggregated residual frame for the current frame; and encode the respective residual blocks in a compressed bitstream. To generate the aggregated residual frame includes to select, for predicting each block of the blocks, a respective selected GMM, where the respective selected GMM corresponds to the one of the at least one respective candidate GMMs that minimizes a total error associated with the aggregated residual frame; and obtain respective residual blocks for the block.

Diversified motion using multiple global motion models

A method for encoding a current frame of a video includes jointly determining respective motion models for reference frames and encoding the current frame using the respective motion models. The reference frames are used for encoding the current frame. Jointly determining respective motion models for reference frames includes determining respective aggregated residuals for combinations of candidate motion models and selecting the combination of candidate motion models that corresponds to the smallest aggregated residual. The respective motion models correspond to the candidate motion models of the selected combination.

Holographic video microscopy yields fast and accurate measurements of the size and refractive index of individual colloidal particles. Particle-resolved characterization offers useful insights into the progress of colloidal synthesis without relying on models for the distributions of particle sizes and properties, and can be performed rapidly enough to provide feedback for process control. The measured increase in the most probable radius over the course of the reaction is consistent with the LaMer model for colloidal growth. Uniformity in the measured refractive index suggests that the spheres grow with uniform density. The joint distribution of size and refractive index provides evidence for a low rate of nucleation proceeding after the initial nucleation event. The same analysis reveals that these PDMS particles shrink by compactification in the first few days after their synthesis.

In-line particle characterization

A system and method for facilitating allocating computing resources to workloads, facilitating workload performance isolation. An example method includes determining one or more workloads to be allocated a set of computing resources in the computing environment, the one or more workloads characterized by metadata describing one or more workload properties; and using the one or more workload properties to calculate a binding between each of the one or more workloads and one or more corresponding portions of the computing resources. Plural competing workloads may be isolated by binding each workload to a disjunct set of Central Processing Units (CPUs) that share as few common hardware resources as possible given a topology the computing resources. Resource allocation adjustments need not require any reconfiguration of the system or adjustment to already provisioned workloads.

Dynamic performance isolation of competing workloads on CPUs with shared hardware components

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for determining when an application is sufficiently instantiated to be subject to a crawling operation. In one aspect, a method includes instantiating an instance of a native application that generates environment instances for display on a user device within the native application instance; monitoring requests for content sent from the native application instance to serving entities that are external to the native application instance; determining whether each monitored request has been fulfilled; in response to determining that the requests are fulfilled: determining whether each thread in a set of threads of the native application instance is idle, generating a load signal in response to determining each thread in the set of thread of the native application instance is idle, the load signal indicating the native application instance is sufficiently instantiated to be subject to a crawling operation.

Monitoring application loading

Growing community of inventors

Jersey City, NJ, United States of America

Chen Wang