Patents by Inventor Udayan Khurana
Udayan Khurana has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Publication number: 20230177032Abstract: A computer-implemented method according to one embodiment includes identifying a data set and meta information; and augmenting the data set with additional features in response to an automatic analysis of the data set in view of the meta information.Type: ApplicationFiled: December 8, 2021Publication date: June 8, 2023Inventors: Daniel Karl I. Weidele, Lisa Amini, Udayan Khurana, Kavitha Srinivas, Horst Cornelius Samulowitz, Takaaki Tateishi, Carolina Maria Spina, Dakuo Wang, Abel Valente, Arunima Chaudhary, Toshihiro Takahashi
-
Publication number: 20230153634Abstract: A domain of an input dataset is identified and one or more archived domain knowledge features corresponding to the identified domain are identified. One or more user feature definitions for one or more user features defined by a user are inputted. The identified archived domain knowledge features and the user features are processed to generate a set of candidate features for presentation to the user. A selection of a subset of the candidate features is obtained from the user and one or more predictive models are generated based on the selected features.Type: ApplicationFiled: November 14, 2021Publication date: May 18, 2023Inventors: Dakuo Wang, Udayan Khurana, Chuang Gan, Gregory Bramble, Abel Valente, Arunima Chaudhary, Carolina Maria Spina, Micah Smith
-
Patent number: 11599826Abstract: Embodiments relate to a system, program product, and method for employing feature engineering to improve classifier performance. A first machine learning (ML) model with a first learning program is selected. The first selected ML model is operatively associated with a first structured dataset. First features in the first dataset directed at performance of the selected ML model are identified. A second structured dataset is assessed with respect to the identified features in the first dataset, and new features in the second dataset are identified, where the new features are semantically related to the identified features in the first dataset. The first dataset is dynamically augmented with the identified new features in the second dataset. The dynamically augmented first dataset is applied to the selected ML model to subject an embedded learning algorithm of the selected ML model to training using the augmented first dataset.Type: GrantFiled: January 13, 2020Date of Patent: March 7, 2023Assignee: International Business Machines CorporationInventors: Udayan Khurana, Sainyam Galhotra, Oktie Hassanzadeh, Kavitha Srinivas, Horst Cornelius Samulowitz
-
Publication number: 20220366269Abstract: A dataset including features and values associated with the features can be received. Each of the features in the dataset can be mapped to a corresponding node in a knowledge graph based on the concept represented by the corresponding node. The knowledge graph can be traversed to find a candidate node connected to at least one mapped node, the candidate node not being mapped to a feature in the dataset. A concept associated with the candidate node can be identified as a new feature. A machine learning model pipeline can use the features in the dataset and the new feature to select a subset of features for training a machine learning model.Type: ApplicationFiled: May 11, 2021Publication date: November 17, 2022Inventors: Dakuo Wang, Udayan Khurana, Daniel Karl I. Weidele, Arunima Chaudhary, Carolina Maria Spina, Abel Valente, Chuang Gan, Horst Cornelius Samulowitz, Lisa Amini
-
Publication number: 20210216904Abstract: Embodiments relate to a system, program product, and method for employing feature engineering to improve classifier performance. A first machine learning (ML) model with a first learning program is selected. The first selected ML model is operatively associated with a first structured dataset. First features in the first dataset directed at performance of the selected ML model are identified. A second structured dataset is assessed with respect to the identified features in the first dataset, and new features in the second dataset are identified, where the new feature is semantically related to the identified features in the first dataset. The first dataset is dynamically augmented with the identified new features in the second dataset. The dynamically augmented first dataset is applied to the selected ML model to subject an embedded learning algorithm of the selected ML model to training using the augmented first dataset.Type: ApplicationFiled: January 13, 2020Publication date: July 15, 2021Applicant: International Business Machines CorporationInventors: Udayan Khurana, Sainyam Galhotra, Oktie Hassanzadeh, Kavitha Srinivas, Horst Cornelius Samulowitz
-
Patent number: 11048718Abstract: Embodiments for feature engineering by one or more processors are described. A plurality of transformations are applied to a set of features in each of a plurality of datasets. An output of each of the plurality of transformations is a score. For each of the sets of features, selecting those of the plurality of transformations for which said score is above a predetermined threshold. A signal representative of said selection is generated.Type: GrantFiled: August 10, 2017Date of Patent: June 29, 2021Assignee: INTERNATIONAL BUSINESS MACHINES CORPORATIONInventors: Elias Khalil, Udayan Khurana, Fatemeh Nargesian, Horst Cornelius Samulowitz, Deepak S. Turaga
-
Patent number: 10353890Abstract: Embodiments relate to analyzing dataset. A method of analyzing data is provided. The method obtains a description of a dataset. The method automatically generates a plurality of analysis options from the description of the dataset. The method generates a plurality of queries based on the analysis options. The method deploys the queries on the dataset to build a plurality of statistical models from the dataset.Type: GrantFiled: June 19, 2015Date of Patent: July 16, 2019Assignee: INTERNATIONAL BUSINESS MACHINES CORPORATIONInventors: Udayan Khurana, Srinivasan Parthasarathy, Venkata N. Pavuluri, Deepak S. Turaga, Long H. Vu
-
Patent number: 10346393Abstract: Embodiments relate to analyzing dataset. A method of analyzing data is provided. The method obtains a description of a dataset. The method automatically generates a plurality of analysis options from the description of the dataset. The method generates a plurality of queries based on the analysis options. The method deploys the queries on the dataset to build a plurality of statistical models from the dataset.Type: GrantFiled: October 20, 2014Date of Patent: July 9, 2019Assignee: INTERNATIONAL BUSINESS MACHINES CORPORATIONInventors: Udayan Khurana, Srinivasan Parthasarathy, Venkata N. Pavuluri, Deepak S. Turaga, Long H. Vu
-
Publication number: 20190050465Abstract: Embodiments for feature engineering by one or more processors are described. A plurality of transformations are applied to a set of features in each of a plurality of datasets. An output of each of the plurality of transformations is a score. For each of the sets of features, selecting those of the plurality of transformations for which said score is above a predetermined threshold. A signal representative of said selection is generated.Type: ApplicationFiled: August 10, 2017Publication date: February 14, 2019Applicant: INTERNATIONAL BUSINESS MACHINES CORPORATIONInventors: Elias KHALIL, Udayan KHURANA, Fatemeh NARGESIAN, Horst Cornelius SAMULOWITZ, Deepak S. TURAGA
-
Patent number: 9858608Abstract: Query suggestions are provided using a query log including a number of user sessions that comprise training data. The training data includes a sequence of a plurality of sets of queries, some of the sets of queries including query transitions followed by a purchase related event. From a cleaned and normalized query log stationary scores and transition scores of at least some of the plurality of sets are generated. A set of query suggestions is built and similarity scores are computed for at least some of the set of query suggestions to determine whether individual ones of the at least some of the set of query suggestions meet a predetermined assurance level. Those that meet the assurance level are included as elements of the set of query suggestions. The set of query suggestions is mixed and ranked according to a user behavior that is sought to be influenced.Type: GrantFiled: March 31, 2016Date of Patent: January 2, 2018Assignee: eBay Inc.Inventors: Mohammad Al Hasan, Nishith Parikh, Gyanit Singh, Neelakantan Sundaresan, Brian Scott Johnson, Udayan Khurana
-
Publication number: 20160217521Abstract: Query suggestions are provided using a query log including a number of user sessions that comprise training data. The training data includes a sequence of a plurality of sets of queries, some of the sets of queries including query transitions followed by a purchase related event. From a cleaned and normalized query log stationary scores and transition scores of at least some of the plurality of sets are generated. A set of query suggestions is built and similarity scores are computed for at least some of the set of query suggestions to determine whether individual ones of the at least some of the set of query suggestions meet a predetermined assurance level. Those that meet the assurance level are included as elements of the set of query suggestions. The set of query suggestions is mixed and ranked according to a user behavior that is sought to be influenced.Type: ApplicationFiled: March 31, 2016Publication date: July 28, 2016Inventors: Mohammad Al Hasan, Nishith Parikh, Gyanit Singh, Neelakantan Sundaresan, Brian Scott Johnson, Udayan Khurana
-
Patent number: 9323811Abstract: Query suggestions are provided using a query log including a number of user sessions that comprise training data. The training data includes a sequence of a plurality of sets of queries, some of the sets of queries including query transitions followed by a purchase related event. From a cleaned and normalized query log stationary scores and transition scores of at least some of the plurality of sets are generated. A set of query suggestions is built and similarity scores are computed for at least some of the set of query suggestions to determine whether individual ones of the at least some of the set of query suggestions meet a predetermined assurance level. Those that meet the assurance level are included as elements of the set of query suggestions. The set of query suggestions is mixed and ranked according to a user behavior that is sought to be influenced.Type: GrantFiled: January 27, 2015Date of Patent: April 26, 2016Assignee: eBay Inc.Inventors: Mohammad Al Hasan, Nishith Parikh, Gyanit Singh, Neelakantan Sundaresan, Brian Scott Johnson, Udayan Khurana
-
Publication number: 20160110410Abstract: Embodiments relate to analyzing dataset. A method of analyzing data is provided. The method obtains a description of a dataset. The method automatically generates a plurality of analysis options from the description of the dataset. The method generates a plurality of queries based on the analysis options. The method deploys the queries on the dataset to build a plurality of statistical models from the dataset.Type: ApplicationFiled: June 19, 2015Publication date: April 21, 2016Inventors: Udayan Khurana, Srinivasan Parthasarathy, Venkata N. Pavuluri, Deepak S. Turaga, Long H. Vu
-
Publication number: 20160110362Abstract: Embodiments relate to analyzing dataset. A method of analyzing data is provided. The method obtains a description of a dataset. The method automatically generates a plurality of analysis options from the description of the dataset. The method generates a plurality of queries based on the analysis options. The method deploys the queries on the dataset to build a plurality of statistical models from the dataset.Type: ApplicationFiled: October 20, 2014Publication date: April 21, 2016Inventors: Udayan Khurana, Srinivasan Parthasarathy, Venkata N. Pavuluri, Deepak S. Turaga, Long H. Vu
-
Publication number: 20150142827Abstract: Query suggestions are provided using a query log including a number of user sessions that comprise training data. The training data includes a sequence of a plurality of sets of queries, some of the sets of queries including query transitions followed by a purchase related event. From a cleaned and normalized query log stationary scores and transition scores of at least some of the plurality of sets are generated. A set of query suggestions is built and similarity scores are computed for at least some of the set of query suggestions to determine whether individual ones of the at least some of the set of query suggestions meet a predetermined assurance level. Those that meet the assurance level are included as elements of the set of query suggestions. The set of query suggestions is mixed and ranked according to a user behavior that is sought to be influenced.Type: ApplicationFiled: January 27, 2015Publication date: May 21, 2015Inventors: Mohammad Al Hasan, Nishith Parikh, Gyanit Singh, Neelakantan Sundaresan, Brian Scott Johnson, Udayan Khurana
-
Patent number: 8954422Abstract: Providing query suggestions using a query log including a number of user sessions that comprise training data including a sequence of a plurality of sets of queries. Some of the sets of queries include query transitions followed by a purchase related event. The query log is cleaned and normalized. Query log stationary scores and transition scores of at least some of the plurality of sets is generated. A set of query suggestions is built and similarity scores are computed for at least some of the set of query suggestions to determine whether individual ones of the at least some of the set of query suggestions meet a predetermined assurance level. Those that meet the level are included as elements of the set of query suggestions that meet the predetermined assurance level. That set of query suggestions are mixed and ranked in accordance with a user behavior sought to be optimized.Type: GrantFiled: July 28, 2011Date of Patent: February 10, 2015Assignee: eBay Inc.Inventors: Mohammad Al Hasan, Nishith Parikh, Gyanit Singh, Neelakantan Sundaresan, Brian Scott Johnson, Udayan Khurana
-
Patent number: 8364462Abstract: A cross-lingual location search uses a combination of translation and transliteration of query tokens to develop a set of candidate matches for further searching. A query is broken up into individual tokens (e.g. address parts) and a list of transliterations and/or translations for each token is developed. The translated and transliterated results are keyed against a spatial database using both literal database keys and transliterated database keys. Matches from the resulting searches are selected when a spatial overlap, or constraint, occurs among subsequences of the query tokens.Type: GrantFiled: June 25, 2008Date of Patent: January 29, 2013Assignee: Microsoft CorporationInventors: Joseph M. Joy, Tanuja Abhay Joshi, Udayan Khurana, Arumugam Kumaran, Vibhuti Singh Sengar, Tobias W. M. Kellner
-
Publication number: 20120036123Abstract: Methods, articles of manufacture and a system for providing query suggestions using a query log that includes a number of user sessions. The sessions comprise training data including a sequence of a plurality of sets of queries, some of the sets of queries including query transitions followed by a purchase related event. The query log is cleaned and normalized. From the cleaned and normalized query log stationary scores and transition scores of at least some of the plurality of sets is generated. A set of query suggestions is built and similarity scores are computed for at least some of the set of query suggestions to determine whether individual ones of the at least some of the set of query suggestions meet a predetermined assurance level. Those that meet the level are included as elements of the set of query suggestions that meet the predetermined assurance level. The set of query suggestions are mixed and ranked in accordance with a user behavior sought to be optimized.Type: ApplicationFiled: July 28, 2011Publication date: February 9, 2012Inventors: Mohammad Al Hasan, Nishith Parikh, Gyanit Singh, Neelakantan Sundaresan, Brian Scott Johnson, Udayan Khurana
-
Publication number: 20090326914Abstract: A cross-lingual location search uses a combination of translation and transliteration of query tokens to develop a set of candidate matches for further searching. A query is broken up into individual tokens (e.g. address parts) and a list of transliterations and/or translations for each token is developed. The translated and transliterated results are keyed against a spatial database using both literal database keys and transliterated database keys. Matches from the resulting searches are selected when a spatial overlap, or constraint, occurs among subsequences of the query tokens.Type: ApplicationFiled: June 25, 2008Publication date: December 31, 2009Applicant: MICROSOFT CORPORATIONInventors: Joseph M. Joy, Tanuja Abhay Joshi, Udayan Khurana, Arumugam Kumaran, Vibhuti Singh Sengar, Tobias W. M. Kellner
-
Publication number: 20090037403Abstract: A location identification system is described. In various embodiments, the location identification system identifies geographic location information in response to received search queries by processing geographic information to identify spatial or geometric regions, determining region intersection information that identifies spatial relationships between the geometric regions, and building an index of regions of constant attributes by associating intersecting geometric regions. In various embodiments, the location identification system can include a vector database wherein the vector database comprises geometric information including at least (a) spatial information geographically describing items and their locations and (b) textual attributes associated with the items or their locations, and an index of regions of constant attributes wherein the index associates textual attributes with items and their locations so that a proximity of two locations can be identified.Type: ApplicationFiled: July 31, 2007Publication date: February 5, 2009Applicant: Microsoft CorporationInventors: Joseph Joy, Tanuja Joshi, Vibhuti Sengar, Udayan Khurana