Google Inc.

Other

Exosome Diagnostics, Inc.

Coty B.v.

Wei Yu

Quoc V Le

Connie P Wang

Golnaz Ghiasi

Michael Andrew Goodman

Ágoston Weisz

Thang Minh Luong

Yifeng Lu

Amin Ghafouri

Sanil Jain

Vikas Peswani

Rakesh Shivanna

Oscar Akerlund

Evgeny Sluzhaev

Igor Petrovski

Hongrae Lee

Johan Skog

Robert Kitchen

Diana Avram

Vasisht Tadigotla

Dalin Chan

Khyatti Gupta

Komal Singh

Elena Castellanos-Rizaldos

Dominik Grimm

Salvatore J Barone

Ralph Macchio

Johan Karl Olov Skog

Dengyong Zhou

Mingxing Tan

Hieu Hy Pham

Zihang Dai

Hanxiao Liu

Rupali A Kulkarni

James Hurley

Antonietta Corrigan

Yu Qing Du

Sudipto Chakrabortty

Mia Sher

Georg Stoll

Yaguang Li

Michael Valentino

Chen Liang

Xinyun Chen

Apoorv Kulshreshtha

Dustin Tran

Xuan Zhang

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using memory-optimized contrastive learning to train image encoder and text encoder neural networks.

Memory-optimized contrastive learning

The present disclosure relates to methods for detecting and treating bladder cancer in a subject using the gene expression level measurements from exosomes derived from the subject.

Methods for developing urine biomarkers and for detecting bladder cancer

Implementations relate to managing multimedia content that is obtained by large language model(s) (LLM(s)) and/or generated by other generative model(s). Processor(s) of a system can: receive natural language (NL) based input that requests multimedia content, generate a response that is responsive to the NL based input, and cause the response to be rendered. In some implementations, and in generating the response, the processor(s) can process, using a LLM, LLM input to generate LLM output, and determine, based on the LLM output, at least multimedia content to be included in the response. Further, the processor(s) can evaluate the multimedia content to determine whether it should be included in the response. In response to determining that the multimedia content should not be included in the response, the processor(s) can cause the response, including alternative multimedia content or other textual content, to be rendered.

Multimedia content management for large language model(s) and/or other generative model(s)

The present disclosure is directed to systems and methods for performing reading comprehension with machine learning. More specifically, the present disclosure is directed to a Neural Symbolic Reader (example implementations of which may be referred to as NeRd), which includes a reader to encode the passage and question, and a programmer to generate a program for multi-step reasoning. By using operators like span selection, the program can be executed over a natural language text passage to generate an answer to a natural language text question. NeRd is domain-agnostic such that the same neural architecture works for different domains. Further, NeRd is compositional such that complex programs can be generated by compositionally applying the symbolic operators.

Neural symbolic reader

Implementations relate to generating multi-modal response(s) through utilization of large language model(s) (LLM(s)). Processor(s) of a system can: receive natural language (NL) based input, generate a multi-modal response that is responsive to the NL based output, and cause the multi-modal response to be rendered. In some implementations, and in generating the multi-modal response, the processor(s) can process, using a LLM, LLM input (e.g., that includes at least the NL based input) to generate LLM output, and determine, based on the LLM output, textual content for inclusion in the multi-modal response and multimedia content for inclusion in the multi-modal response. In some implementations, the multimedia content can be obtained based on a multimedia content tag that is included in the LLM output and that is indicative of the multimedia content. In various implementations, the multimedia content can be interleaved between segments of the textual content.

Generating multi-modal response(s) through utilization of large language model(s)

The present invention relates generally to methods and kits for detecting one or more biomarkers, such as an Epidermal Growth Factor Receptor (EGFR) mutation, e.g., T790M mutation, L858R mutation, one or more exon19 insertions and/or one or more exon19 deletions in the EGFR gene, in a biological sample to aid in diagnosis, prognosis, monitoring, or therapy selection for a disease such as, for example, cancer. The methods and kits are useful in aiding in diagnosis, prognosis, monitoring, or therapy selection for lung cancer, e.g., non-small cell lung cancer (NSCLC).

Methods and compositions to detect mutations in plasma using exosomal RNA and cell free DNA from non-small cell lung cancer patients

The invention provides a series of steps that prepare nucleic acids (RNA and/or DNA) isolated from extracellular vesicles for sequencing. This enables a wide diversity of RNAs and/or DNAs, to be efficiently detected. These can then be used to identify various attributes such as gene expression, alternative splicing, and the detection of both somatic and germline mutations including single nucleotide variants (SNV) and structural variations (insertions/deletions, fusions, inversions).

Sequencing and analysis of exosome associated nucleic acids

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for processing sequential data. In one aspect, a computer-implemented method includes receiving a request to generate a system output for an input data sequence, the input data sequence including a plurality of tokens. One or more tokens may be designated as tokens to be skipped. When a token has not been designated as a token to be skipped, the token is processed using a recurrent neural network to update a current internal state of the recurrent neural network. The system output is generated from the final internal state of the recurrent neural network.

Skimming data sequences using recurrent neural networks

Skimming text using recurrent neural networks

This invention provides a cup with twinkling light effects, primarily comprising an inner cup, an outer cup, and a light unit, with the lower half of the inner cup having a number of arched convex strips, and the base of which has diamond-facet of concave and convex cones. As a result, when the light unit at the base of the outer cup is activated, the effects of the arched convex strips and diamond facets cause the cup to produce numerous refractions of the light source, creating twinkling rays of light with dazzling variation.

Cup with twinkling light effects

Luminous stick

The present invention relates to a straw cup with variable lights, primarily comprising of a set of light-emitting diodes surrounding the inside of the cup's straw, a circuit controller board, and power switch; when the power is activated, the liquid inside the straw's cup refracts and directs light, increasing the variation and brilliance of the twinkling lights and thus further increasing the variation of the straw cup's multicolored lights.

Straw cup with variable lights

This invention is a light baton for use in aquatic games, comprising a translucent baton body, end inserts at both ends of the baton body, an end insert connected to an inserted light-emitting module, holes in the translucent baton body, and arcuate convex protuberances on the outer surface of the translucent baton body; the holes in the translucent baton body allow water to flow in and surround the light-emitting module, with the change in air and water density causing the baton to sink in water, while the liquid has a refractive effect on the colorful lights from the light-emitting module; the colorful beams of light pass through and are refracted by the ripples in the liquid within the baton body to produce illumination, making the baton ideal for having fun in water.

Light baton for use in aquatic games

A new personal care product in the form of a free-standing stick comprises an inner core with cosmetic ingredients and an solid outer shell adhering to the inner core, in which the outer shell is wax-free and contains at least solid proteins, liquid polyols or solutions of polyols, solid polyols, emollients and additional cosmetic auxiliary materials and active ingredients. The outer surface does not transfer pigment through brief contact with the skin and is also not sticky.

Personal care product in the form of a free-standing stick

Growing community of inventors

Pittsburgh, PA, United States of America

Wei Yu