Li Yang

8030354, Oct 4, 2011

Oct 3, 2008

12/287036

Sean P. Brown - San Francisco CA, US
Qiong Cao - Daly City CA, US
Paul John Dransfield - San Francisco CA, US
Xiaohui Du - Belmont CA, US
Zice Fu - Foster City CA, US
Jonathan Houze - San Mateo CA, US
Xian Yun Jiao - Belmont CA, US
Yong-Jae Kim - Foster City CA, US
Todd J. Kohn - San Mateo CA, US
SuJen Lai - Burlingame CA, US
Daniel Lin - Redwood City CA, US
Jiwen Liu - Foster City CA, US
Jian Luo - Albany CA, US
Julio C. Medina - San Carlos CA, US
Jeffrey D. Reagan - Woodside CA, US
Vatee Pattaropong - Burlingame CA, US
Margrit Schwarz - San Carlos CA, US
Wang Shen - San Mateo CA, US
Yongli Su - Foster City CA, US
Gayathri Swaminath - Fremont CA, US
Marc Vimolratana - San Mateo CA, US
Yingcai Wang - Millbrae CA, US
Yumei Xiong - San Bruno CA, US
Li Yang - San Francisco CA, US
Ming Yu - Foster City CA, US
Jie Zhang - Sunnyvale CA, US
Liusheng Zhu - Foster City CA, US

Amgen Inc. - Thousand Oaks CA

A61K 31/192
C07C 53/134

514570, 562496, 562512, 546342, 514357

The present invention provides compounds useful, for example, for treating metabolic disorders in a subject. Such compounds have the general formula I:.

20100262499, Oct 14, 2010

Apr 10, 2009

12/385554

Niklas Karlsson - Mountain VIew CA, US
Li Yang - Mountain View CA, US

G06Q 30/00
G06Q 99/00

705 1472, 705400

A system is provided for controlling initialization of an online advertising campaign. The system includes a sensor configured to determine current performance of the online advertising campaign and a memory for storing performance information obtained by the sensor, a set of campaign initialization controller instructions, a campaign volume model, and a network average time-of-day pattern. The system also includes an estimator configured to predict future campaign performance based on the performance information obtained by the sensor, the campaign volume model, and the network average time-of-day pattern. The system also includes an initialization controller configured to generate a bid allocation control signal and generate a bid price control signal based on the future campaign performance predicted by the estimator, according to the instructions stored in the memory. A method for controlling initialization of an online advertising campaign is also provided.

20230078473, Mar 16, 2023

Sep 14, 2022

17/932104

Deliang Fan - Tempe AZ, US
Adnan Siraj Rakin - Tempe AZ, US
Li Yang - Tempe AZ, US
Chaitali Chakrabarti - Tempe AZ, US
Yu Cao - Gilbert AZ, US
Jae-sun Seo - Tempe AZ, US
Jingtao Li - Tempe AZ, US

Arizona Board of Regents on behalf of Arizona State University - Scottsdale AZ

G06N 3/08
G06N 3/04

A robust and accurate binary neural network, referred to as RA-BNN, is provided to simultaneously defend against adversarial noise injection and improve accuracy. Recently developed adversarial weight attack, a.k.a. bit-flip attack (BFA), has shown enormous success in compromising deep neural network (DNN) performance with an extremely small amount of model parameter perturbation. To defend against this threat, embodiments of RA-BNN adopt a complete binary neural network (BNN) to significantly improve DNN model robustness (defined as the number of bit-flips required to degrade the accuracy to as low as a random guess). To improve clean inference accuracy, a novel and efficient two-stage network growing method is proposed and referred to as early growth. Early growth selectively grows the channel size of each BNN layer based on channel-wise binary masks training with Gumbel-Sigmoid function. Apart from recovering the inference accuracy, the RA-BNN after growing also shows significantly higher resistance to BFA.

20210304457, Sep 30, 2021

Feb 19, 2021

17/180182

- Oakland CA, US
- Otawara-shi, JP
Zhaoheng XIE - Oakland CA, US
Wenyuan QI - Vernon Hills IL, US
Li YANG - Vernon Hills IL, US
Chung CHAN - Vernon Hills IL, US
Evren ASMA - Vernon Hills IL, US

The Regents of the University of California - Oakland CA
CANON MEDICAL SYSTEMS CORPORATION - Otawara-shi

G06T 11/00
G06T 7/246

To reduce the effect(s) caused by patient breathing and movement during PET data acquisition, an unsupervised non-rigid image registration framework using deep learning is used to produce motion vectors for motion correction. In one embodiment, a differentiable spatial transformer layer is used to warp the moving image to the fixed image and use a stacked structure for deformation field refinement. Estimated deformation fields can be incorporated into an iterative image reconstruction process to perform motion compensated PET image reconstruction. The described method and system, using simulation and clinical data, provide reduced error compared to at least one iterative image registration process.

20200403958, Dec 24, 2020

Oct 31, 2018

16/177263

- Menlo Park CA, US
Li Yang - Palo Alto CA, US
Henri Romeo Liriani - San Francisco CA, US
Tao Jiang - Newark CA, US
Xu He - Fremont CA, US

H04L 12/58

Systems, methods, and non-transitory computer readable media can determine information associated with a web page being accessed by a user operating a computing device. The information associated with the web page can be verified. A messaging interface for exchanging messages with an entity associated with the web page can be provided. The messaging interface can be provided within the web page being accessed.

20200314201, Oct 1, 2020

Mar 29, 2019

16/370690

- Redmond WA, US
Li Yang - Palo Alto CA, US
Yongzheng Zhang - San Jose CA, US
Shen Huang - San Jose CA, US
Clayton Sanford - San Francisco CA, US

H04L 29/08
G06N 20/00
H04L 12/911
G09B 5/00

An embodiment of the disclosed technologies includes extracting, from an online connection network, digital data comprising target profile data and current profile data; where the target profile data is associated with an online submission process that has a plurality of possible outcomes and is executable via the online connection network; where the current profile data is associated with a member node of the online connection network; using an active learning process, in response to the current profile data, identifying attribute data that is in the target profile data but is not in the current profile data and is predicted to have a relationship with a positive outcome of the online submission process; outputting the attribute data for use by a downstream process or an automated digital assistant to determine a digital resource to associate with the member node through the online connection network or through an online learning system.

20200101388, Apr 2, 2020

Dec 3, 2019

16/702253

- Menlo Park CA, US
Richard Kenneth Zadorozny - Menlo Park CA, US
Jeremy Harrison Goldberg - San Francisco CA, US
Peter Henry Martinazzi - San Francisco CA, US
Lei Guang - Bellevue WA, US
Li Yang - Palo Alto CA, US
Jennifer A. Burge - Oakland CA, US

A63F 13/87
G06Q 50/00
A63F 13/79
A63F 13/2145
A63F 13/48
A63F 13/92
A63F 13/35
G06F 21/10
A63F 13/795
A63F 13/49

This disclosure is directed to enabling access to a game instance through a messenger application. A client-side player component facilitates communication between a game instance on the client device and a backend server of a social networking system. The communication information can be stored in or retrieved from a social graph of the social networking system. The game instance can be initialized from a social platform of the social networking system, and the game can be played without leaving the messenger application.

20200106804, Apr 2, 2020

Sep 27, 2018

16/145032

- Redmond WA, US
Kaiyu Yang - Sunnyvale CA, US
Li Yang - Palo Alto CA, US
Bing Wang - Mountain View CA, US

H04L 29/06
G06N 99/00
H04L 12/58

A spammer profile detector uses multi-stage machine learning approach, where a content-based machine learning model, a connection graph machine learning model, and a behavior-based machine learning model are used sequentially, each model generating a score indicating the likelihood that a profile is a spammer profile. The content-based machine learning model examines and evaluates information stored in a member profile. The connection graph machine learning model examines and evaluates a member's connections. The behavior-based machine learning model examines and evaluates activities of a member represented by a member profile. The score produced by the spammer profile detector can be used to determine whether the profile should be flagged as a spammer profile, whether the profile should be omitted when determining a count of the total number of active member profiles within the system, whether the profile should be restricted or removed from the system, etc.