Yan J Liu

US Patent:

7277574, Oct 2, 2007

Filed:

Jun 27, 2005

Appl. No.:

11/169107

Inventors:

Yan Liu - New York NY, US
John Kender - Leonia NJ, US

Assignee:

The Trustees of Columbia University in the City of New York - New York NY

International Classification:

G06K 9/62

US Classification:

382159, 707 7, 712300

Abstract:

Methods and systems for feature selection are described. In particular, methods and systems for feature selection for data classification, retrieval, and segmentation are described. Certain embodiments of the invention are directed to methods and systems for complement sort-merge tree (CSMT), fast-converging sort-merge tree (FSMT), and multi-level (ML) feature selection. Accurate and fast results may be obtained by the feature selection methods and systems described herein.

US Patent:

8255346, Aug 28, 2012

Filed:

Nov 11, 2009

Appl. No.:

12/616534

Inventors:

Naoki Abe - Yorktown Heights NY, US
Yan Liu - Yorktown Heights NY, US
Aurelie C. Lozano - Yorktown Heights NY, US
Saharon Rosset - Tel Aviv, IL
Grzegorz Swirszcz - Yorktown Heights NY, US

Assignee:

International Business Machines Corporation - Armonk NY

International Classification:

G06F 15/18
G06F 7/60
G06F 17/50

US Classification:

706 15, 706 19, 706 25, 703 2, 703 13

Abstract:

A “variable group selection” system and method in which constructs are based upon a training data set, a regression modeling module that takes into account information on groups of related predictor variables given as input and outputs a regression model with selected variable groups. Optionally, the method can be employed as a component in methods of temporal causal modeling, which are applied on a time series training data set, and output a model of causal relationships between the multiple times series in the data.

US Patent:

8332379, Dec 11, 2012

Filed:

Jun 11, 2010

Appl. No.:

12/813608

Inventors:

Ning Duan - Beijing, CN
Pei-Yun S. Hsueh - White Plains NY, US
Yan Liu - Elmsford NY, US

Assignee:

International Business Machines Corporation - Armonk NY

International Classification:

G06F 17/30

US Classification:

707706, 707707, 707708, 707709, 707723, 707748, 707749, 707750, 707751, 707753

Abstract:

A method and system for identifying nodes with similar content. In one aspect, the method comprises determining a structure of a network of nodes, said structure defined by incoming links and outgoing links between nodes within said network, grouping said nodes within said network into a first set of modules, calculating a first modularity value between each of the modules within the first set, said modularity value indicating a degree of similar content within each module, calculating a topical relevance value for each of the modules, selecting those modules whose topical relevance value exceeds a threshold value and calculating an authority score for the selected modules.

US Patent:

20030167505, Sep 4, 2003

Filed:

Mar 11, 2002

Appl. No.:

10/094778

Inventors:

Brian Dilkes - Tucson AZ, US
William Gordon-Kamm - Urbandale IA, US
Paolo Sabelli - Tucson AZ, US
Brian Larkins - Tucson AZ, US
Keith Lowe - Johnston IA, US
Yuejin Sun - Westfield IN, US
Chunsheng Zhang - North Charleston SC, US
Yan Liu - Corona NY, US

International Classification:

A01H005/00
C07H021/04
C12N005/04

US Classification:

800/278000, 435/419000, 435/320100, 536/023600

Abstract:

The invention provides isolated Rb3 nucleic acids and their encoded proteins that are involved in cell cycle regulation. The invention further provides recombinant expression cassettes, host cells, transgenic plants, and antibody compositions. The present invention provides methods and compositions relating to altering cell cycle protein content, cell cycle progression, cell number and/or composition of plants.

US Patent:

20110004463, Jan 6, 2011

Filed:

Jul 1, 2009

Appl. No.:

12/496252

Inventors:

Wojciech Gryc - Ontario, CA
Richard Lawrence - Yorktown Heights NY, US
Yan Liu - Yorktown Heights NY, US

Assignee:

INTERNATIONAL BUSINESS MACHINES CORPORATION - Armonk NY

International Classification:

G06F 17/27

US Classification:

704 9

Abstract:

A computing system receives input data having both graph and unstructured data and computes a current log likelihood of the input data. The computing system compares the current log likelihood with a previous log likelihood of the input data. If the current log likelihood is larger than the previous log likelihood, the computing system update topic modeling parameters, community modeling parameters, and the link generation parameter until the computing system obtains a maximal value of the log likelihood of the input data. Then, the computing system creates a graph indicating topic similarity between the input data based on the topic modeling parameters, creates another graph indicating community similarity between entities associated with the input data based on the community modeling parameters, and predicts a link existence between input data or entities based on the link generation parameter, the topic modeling parameter and the community modeling parameter.

US Patent:

20110111384, May 12, 2011

Filed:

Nov 6, 2009

Appl. No.:

12/614175

Inventors:

Brenda Dietrich - Yorktown Heights NY, US
Jennifer Lai - Garrison NY, US
Yan Liu - Elmsford NY, US

Assignee:

INTERNATIONAL BUSINESS MACHINES CORPORATION - Armonk NY

International Classification:

G09B 7/00
G09B 19/00

US Classification:

434350, 434238, 434362

Abstract:

A method and system for capturing expert behavior that the expert is not even necessarily conscious of, such as gaze patterns, that are tied to positive outcomes, and to create a catalog (e.g. database) of these behaviors. The system compares the current behavior of a person, for example a novice, against this catalog of expert behaviors. If the current behavior is significantly different, the system would adapt various aspects of the interface to bring the current behavior closer in line to that of the expert.

US Patent:

20110320387, Dec 29, 2011

Filed:

Nov 2, 2010

Appl. No.:

12/938063

Inventors:

Jingrui He - Ossining NY, US
Richard D. Lawrence - Ridgefield CT, US
Yan Liu - Elmsford NY, US

Assignee:

INTERNATIONAL BUSINESS MACHINES CORPORATION - Armonk NY

International Classification:

G06F 15/18

US Classification:

706 12

Abstract:

Transfer learning is the task of leveraging the information from labeled examples in some domains to predict the labels for examples in another domain. It finds abundant practical applications, such as sentiment prediction, image classification and network intrusion detection. A graph-based transfer learning framework propagates label information from a source domain to a target domain via the example-feature-example tripartite graph, and puts more emphasis on the labeled examples from the target domain via the example-example bipartite graph. An iterative algorithm renders the framework scalable to large-scale applications. The framework propagates the label information to both features irrelevant to the source domain and unlabeled examples in the target domain via common features in a principled way.

US Patent:

20120185415, Jul 19, 2012

Filed:

Jan 13, 2011

Appl. No.:

13/006245

Inventors:

Vijil E. Chenthamarakshan - Kerala, IN
Richard D. Lawrence - Ridgefield CT, US
Yan Liu - Elmsford NY, US
Dan Zhang - West Lafayette IN, US

Assignee:

INTERNATIONAL BUSINESS MACHINES CORPORATION - Armonk NY

International Classification:

G06F 15/18

US Classification:

706 12

Abstract:

System, method and computer program product provides a novel domain adaption/transfer learning approach applied to the problem of classifying abbreviated documents, e.g., short text messages, instant messages, tweets. The proposed method uses a large number of multi-labeled examples (source domain) to improve the learning on the partial observations (target domain). Specifically, a hidden, higher-level abstraction space is learned that is meaningful for the multi-labeled examples in the source domain. This is done by simultaneously minimizing the document reconstruction error and the error in a classification model learned in the hidden space using known labels from the source domain. The partial observations in the target space are then mapped to the same hidden space, and classified into the label space determined by the source domain. Exemplary results provided for a Twitter dataset demonstrate that the method identifies meaningful hidden topics and provides useful classifications of specific tweets.