Lihong H Li, 4411511 159Th Ave NE, Redmond, WA 98052

2011013, Jun 2, 2011

Nov 30, 2009

12/627617

Jason Williams - New York NY, US
Suhrid Balakrishnan - Westfield NJ, US
Lihong Li - Santa Clara CA, US

AT&T Intellectual Property I, L.P. - Reno NV

G10L 21/00

704270, 704E11001

Disclosed herein are systems, methods, and computer-readable storage media for automatically generating a dialog manager for use in a spoken dialog system. A system practicing the method receives a set of user interactions having features, identifies an initial policy, evaluates all of the features in a linear evaluation step of the algorithm to identify a set of most important features, performs a cubic policy improvement step on the identified set of most important features, repeats the previous two steps one or more times, and generates a dialog manager for use in a spoken dialog system based on the resulting policy and/or set of most important features. Evaluating all of the features can include estimating a weight for each feature which indicates how much each feature contributes to at least one of the identified policies. The system can ignore features not in the set of most important features.

2012001, Jan 19, 2012

Jul 14, 2010

12/836188

Lihong Li - Santa Clara CA, US
Wei Chu - San Jose CA, US
John Langford - White Plains NY, US
Robert Schapire - Princeton NJ, US

YAHOO! INC. - Sunnyvale CA

G06F 17/10
G06F 15/173

703 2, 709224

Methods and apparatus for performing computer-implemented personalized recommendations are disclosed. User information pertaining to a plurality of features of a plurality of users may be obtained. In addition, item information pertaining to a plurality of features of the plurality of items may be obtained. A plurality of sets of coefficients of a linear model may be obtained based at least in part on the user information and/or the item information such that each of the plurality of sets of coefficients corresponds to a different one of a plurality of items, where each of the plurality of sets of coefficients includes a plurality of coefficients, each of the plurality of coefficients corresponding to one of the plurality of features. In addition, at least one of the plurality of coefficients may be shared among the plurality of sets of coefficients for the plurality of items. Each of a plurality of scores for a user may be calculated using the linear model based at least in part upon a corresponding one of the plurality of sets of coefficients associated with a corresponding one of the plurality of items, where each of the plurality of scores indicates a level of interest in a corresponding one of a plurality of items. A plurality of confidence intervals may be ascertained, each of the plurality of confidence intervals indicating a range representing a level of confidence in a corresponding one of the plurality of scores associated with a corresponding one of the plurality of items. One of the plurality of items for which a sum of a corresponding one of the plurality of scores and a corresponding one of the plurality of confidence intervals is highest may be recommended.

2013011, May 2, 2013

Oct 26, 2011

13/282285

Wei Chu - Redmond WA, US
Martin Zinkevich - Santa Clara CA, US
Lihong Li - Santa Clara CA, US
Achint Oommen Thomas - Buffalo NY, US
Belle Tseng - Cupertino CA, US

Yahoo!, Inc. - Sunnyvale CA

G06F 15/16
G06N 5/02
G06F 11/00

709224

Software for online active learning receives content posted to an online stream at a website. The software converts the content into an elemental representation and inputs the elemental representation into a probit model to obtain a predictive probability that the content is abusive. The software also calculates an importance weight based on the elemental representation. And the software updates the probit model using the content, the importance weight, and an acquired label if a condition is met. The condition depends on an instrumental distribution. The software removes the content from the online stream if a condition is met. The condition depends on the predictive probability, if an acquired label is unavailable.

2013023, Sep 12, 2013

Apr 30, 2013

13/873661

Suhrid Balakrishnan - Westfield NJ, US
Lihong Li - Santa Clara CA, US

AT&T Intellectual Property I, L.P. - Atlanta GA

G10L 15/06

704244

Disclosed herein are systems, methods, and computer-readable storage media for automatically generating a dialog manager for use in a spoken dialog system. A system practicing the method receives a set of user interactions having features, identifies an initial policy, evaluates all of the features in a linear evaluation step of the algorithm to identify a set of most important features, performs a cubic policy improvement step on the identified set of most important features, repeats the previous two steps one or more times, and generates a dialog manager for use in a spoken dialog system based on the resulting policy and/or set of most important features. Evaluating all of the features can include estimating a weight for each feature which indicates how much each feature contributes to at least one of the identified policies. The system can ignore features not in the set of most important features.

2019032, Oct 24, 2019

Apr 24, 2018

15/960809

- Redmond WA, US
Xiujun LI - Bellevue WA, US
Lihong LI - Redmond WA, US
Da TANG - New York NY, US
Chong WANG - Bellevue WA, US
Tony JEBARA - Los Gatos CA, US

Microsoft Technology Licensing, LLC - Redmond WA

G06F 9/48
G06F 9/50
G06F 17/30
G06F 17/28
G06N 3/02
G10L 15/22

A system for executing composite tasks can include a processor to detect a composite task from a user. The processor can also detect a plurality of subtasks corresponding to the composite task based on unsupervised data without a label, wherein the plurality of subtasks are identified by a top-level dialog policy. The processor can also detect a plurality of actions, wherein each action is to complete one of the subtasks, and wherein each action is identified by a low-level dialog policy corresponding to the subtasks identified by the top-level dialog policy. The processor can also update a dialog manager based on a completion of each action corresponding to the subtasks and execute instructions based on a policy identified by the dialog manager, wherein the executed instructions implement the policy with a lowest global cost corresponding to the composite task provided by the user.

2019000, Jan 3, 2019

Jun 30, 2017

15/639304

- Redmond WA, US
Li Deng - Redmond WA, US
Lihong Li - Redmond WA, US
Chong Wang - Redmond WA, US

G06N 3/08
G06N 3/00

Systems and methods are disclosed for inquiry-based deep learning. In one implementation, a first content segment is selected from a body of content. The content segment includes a first content element. The first content segment is compared to a second content segment to identify a content element present in the first content segment that is not present in the second content segment. Based on an identification of the content element present in the first content segment that is not present in the second content segment, the content element is stored in a session memory. A first question is generated based on the first content segment. The session memory is processed to compute an answer to the first question. An action is initiated based on the answer. Using deep learning, content segments can be encoded into memory. Incremental questioning can serve to focus various deep learning operations on certain content segments.

2018027, Sep 27, 2018

Mar 27, 2017

15/470784

- Redmond WA, US
Rishabh Singh - Kirkland WA, US
Lihong Li - Redmond WA, US
Pushmeet Kohli - Bellevue WA, US
Emilio Parisotto - Pittsburgh PA, US

G06F 9/44
G06N 3/02

Described are systems, methods, and computer-readable media for program generation in a domain-specific language based on input-output examples. In accordance with various embodiments, a neural-network-based program generation model conditioned on an encoded set of input-output examples is used to generate a program tree by iteratively expanding a partial program tree, beginning with a root node and ending when all leaf nodes are terminal.

2018025, Sep 6, 2018

May 7, 2018

15/973130

- New York NY, US
Martin Zinkevich - Santa Clara CA, US
Lihong Li - Santa Clara CA, US
Achint Oommen Thomas - Buffalo NY, US
Belle Tseng - Cupertino CA, US

H04L 12/58
G06N 5/02
G06F 11/00
G06Q 50/20
G06N 7/00
G06F 15/16

Software for online active learning receives content posted to an online stream at a website. The software converts the content into an elemental representation and inputs the elemental representation into a probit model to obtain a predictive probability that the content is abusive. The software also calculates an importance weight based on the elemental representation. And the software updates the probit model using the content, the importance weight, and an acquired label if a condition is met. The condition depends on an instrumental distribution. The software removes the content from the online stream if a condition is met. The condition depends on the predictive probability, if an acquired label is unavailable.