Patents by Inventor Mahesh A. Makhijani
Mahesh A. Makhijani 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).
-
Patent number: 11783375Abstract: Provided is a process, including: obtaining a first training dataset, training a first machine-learning model on the first training dataset, obtaining a set of candidate question sequences, forming virtual subject-entity records, forming a second training dataset, training a second machine-learning model, and storing the adjusted parameters of the second machine-learning model in memory.Type: GrantFiled: July 8, 2022Date of Patent: October 10, 2023Assignee: Cerebri AI Inc.Inventors: Alain Charles Briancon, Jean Joseph Belanger, James Cvetan Stojanov, Christopher Michael Coovrey, Pranav Mahesh Makhijani, Gregory Klose, Max Changchun Huang, Mounib Mohamad Ismail, Michael Henry Engeling, Hongshi Li
-
Publication number: 20230101487Abstract: Provided is a process, including: obtaining a first training dataset, training a first machine-learning model on the first training dataset, obtaining a set of candidate question sequences, forming virtual subject-entity records, forming a second training dataset, training a second machine-learning model, and storing the adjusted parameters of the second machine-learning model in memory.Type: ApplicationFiled: July 8, 2022Publication date: March 30, 2023Inventors: Alain Charles Briancon, Jean Joseph Belanger, James Cvetan Stojanov, Christopher Michael Coovrey, Pranav Mahesh Makhijani, Gregory Klose, Max Changchun Huang, Mounib Mohamad Ismail, Michael Henry Engeling, Hongshi Li
-
Patent number: 11615271Abstract: Provided is a process of modeling methods organized in racks of a machine learning pipeline to facilitate optimization of performance using modelling methods for implementation of machine learning design in an object-oriented modeling (OOM) framework, the process including: writing classes using object-oriented modelling of optimization methods, modelling methods, and modelling racks; writing parameters and hyper-parameters of the modeling methods as attributes as the modeling methods; scanning modelling racks classes to determine first class definition information; selecting a collection of rack and selecting modeling method objects; scanning modelling method classes to determine second class definition information; assigning racks and locations within the racks to modeling method objects; and invoking the class definition information to produce object manipulation functions that allow access the methods and attributes of at least some of the modeling method objects, the manipulation functions being configurType: GrantFiled: June 3, 2020Date of Patent: March 28, 2023Assignee: Cerebri AI Inc.Inventors: Eyal Ben Zion, Alain Charles Briancon, Pranav Mahesh Makhijani, Thejas Narayana Prasad, Sara Amini, Jian Deng, Ngoc Thu Nguyen, Jean Joseph Belanger
-
Patent number: 11416896Abstract: Provided is a process, including: obtaining a first training dataset, training a first machine-learning model on the first training dataset, obtaining a set of candidate question sequences, forming virtual subject-entity records, forming a second training dataset, training a second machine-learning model, and storing the adjusted parameters of the second machine-learning model in memory.Type: GrantFiled: June 18, 2021Date of Patent: August 16, 2022Assignee: Cerebri AI Inc.Inventors: Alain Charles Briancon, Jean Joseph Belanger, James Cvetan Stojanov, Christopher Michael Coovrey, Pranav Mahesh Makhijani, Gregory Klose, Max Changchun Huang, Mounib Mohamad Ismail, Michael Henry Engeling, Hongshi Li
-
Publication number: 20220044283Abstract: Provided is a process, including: obtaining a first training dataset, training a first machine-learning model on the first training dataset, obtaining a set of candidate question sequences, forming virtual subject-entity records, forming a second training dataset, training a second machine-learning model, and storing the adjusted parameters of the second machine-learning model in memory.Type: ApplicationFiled: June 18, 2021Publication date: February 10, 2022Inventors: Alain Charles Briancon, Jean Joseph Belanger, James Cvetan Stojanov, Christopher Michael Coovrey, Pranav Mahesh Makhijani, Gregory Klose, Max Changchun Huang, Mounib Mohamad Ismail, Michael Henry Engeling, Hongshi Li
-
Patent number: 11068942Abstract: Provided is a process, including: obtaining a first training dataset, training a first machine-learning model on the first training dataset, obtaining a set of candidate question sequences, forming virtual subject-entity records, forming a second training dataset, training a second machine-learning model, and storing the adjusted parameters of the second machine-learning model in memory.Type: GrantFiled: October 18, 2019Date of Patent: July 20, 2021Assignee: Cerebri AI Inc.Inventors: Alain Charles Briancon, Jean Joseph Belanger, James Cvetan Stojanov, Christopher Michael Coovrey, Pranav Mahesh Makhijani, Gregory Klose, Max Changchun Huang, Mounib Mohamad Ismail, Michael Henry Engeling, Hongshi Li
-
Publication number: 20200380416Abstract: Provided is a process of modeling methods organized in racks of a machine learning pipeline to facilitate optimization of performance using modelling methods for implementation of machine learning design in an object-oriented modeling (OOM) framework, the process including: writing classes using object-oriented modelling of optimization methods, modelling methods, and modelling racks; writing parameters and hyper-parameters of the modeling methods as attributes as the modeling methods; scanning modelling racks classes to determine first class definition information; selecting a collection of rack and selecting modeling method objects; scanning modelling method classes to determine second class definition information; assigning racks and locations within the racks to modeling method objects; and invoking the class definition information to produce object manipulation functions that allow access the methods and attributes of at least some of the modeling method objects, the manipulation functions being configurType: ApplicationFiled: June 3, 2020Publication date: December 3, 2020Inventors: Eyal Ben Zion, Alain Charles Briancon, Pranav Mahesh Makhijani, Thejas Narayana Prasad, Sara Amini, Jian Deng, Ngoc Thu Nguyen, Jean Joseph Belanger
-
Publication number: 20200126126Abstract: Provided is a process, including: obtaining a first training dataset, training a first machine-learning model on the first training dataset, obtaining a set of candidate question sequences, forming virtual subject-entity records, forming a second training dataset, training a second machine-learning model, and storing the adjusted parameters of the second machine-learning model in memory.Type: ApplicationFiled: October 18, 2019Publication date: April 23, 2020Inventors: Alain Charles Briancon, Jean Joseph Belanger, James Cvetan Stojanov, Christopher Michael Coovrey, Pranav Mahesh Makhijani, Gregory Klose, Max Changchun Huang, Mounib Mohamad Ismail, Michael Henry Engeling, Hongshi Li
-
Patent number: 8902806Abstract: A method of managing macro network coverage among a plurality of repeaters deployed to extend coverage of one or more base stations may include receiving feedback information from at least two repeaters indicative of macro network coverage at locations corresponding to each repeater, determining a coverage status of the macro network based on the feedback information, and sending a control signal to at least one repeater instructing the repeater to adjust one or more operating parameters based on the determined coverage status.Type: GrantFiled: March 22, 2011Date of Patent: December 2, 2014Assignee: QUALCOMM IncorporatedInventors: Eduardo S. Esteves, Dhananjay Ashok Gore, Mahesh A. Makhijani
-
Patent number: 8576931Abstract: Methods and apparatus are provided for increasing throughput in a wireless communication system by reducing the amount of overhead transmitted to certain user terminals. Overhead due to control information may be reduced for these certain user terminals by selecting a low repetition factor. Overhead may be further reduced for these certain user terminals by selecting a modulation/coding scheme with a higher data rate for transmitting the control information. The selection may be based on channel conditions associated with the user terminals, such as signal-to-interference-plus-noise ratios (SINRs).Type: GrantFiled: June 17, 2009Date of Patent: November 5, 2013Assignee: QUALCOMM IncorporatedInventors: Serguei A. Glazko, Matthias Brehler, Je Woo Kim, Mahesh Makhijani
-
Patent number: 8233452Abstract: Techniques for sending and receiving signaling messages on shared and dedicated channels in a wireless communication system are described. A user equipment (UE) may receive a first signaling message on a shared channel during a first time period, receive a second signaling message on a dedicated channel during a second time period, and receive a third signaling message on the shared channel during a third time period. The UE may receive traffic data on the shared channel during all time periods. The UE may be handed over from a source Node B to a target Node B during the second time period. The UE may receive the first signaling message via the source Node B, receive the second signaling message via the source and target Node Bs, and receive the third signaling message via the target Node B.Type: GrantFiled: September 15, 2008Date of Patent: July 31, 2012Assignee: QUALCOMM IncorporatedInventors: Mahesh Makhijani, Yavuz Mehmet
-
Publication number: 20110235568Abstract: A method of managing macro network coverage among a plurality of repeaters deployed to extend coverage of one or more base stations may include receiving feedback information from at least two repeaters indicative of macro network coverage at locations corresponding to each repeater, determining a coverage status of the macro network based on the feedback information, and sending a control signal to at least one repeater instructing the repeater to adjust one or more operating parameters based on the determined coverage status.Type: ApplicationFiled: March 22, 2011Publication date: September 29, 2011Applicant: QUAL COMM IncorporatedInventors: Eduardo S. Esteves, Dhananjay Ashok Gore, Mahesh A. Makhijani
-
Patent number: 7742455Abstract: A scheduler at a base station may schedule packet data traffic based on a ranking metric that varies directly with the mobile station's scheduling downlink transmission rate and a delay factor indicative of the staleness of the corresponding queued data. The ranking metric may advantageously vary in a direct non-linear fashion with the delay factor to allow for delay sensitive data, such as VoIP data, to be scheduled with increased urgency when quality of service is about to be compromised. The scheduler may attempt to pack a multi-user downlink physical layer packet by selecting a tentative rate and determining if an aggregate amount of data in the packet may be increased by transmitting the packet at a lower rate. If so, additional queued data is added to the packet and the transmission rate for the packet is lowered. Such an approach allows for greater link efficiency to be achieved.Type: GrantFiled: October 13, 2005Date of Patent: June 22, 2010Assignee: Telefonaktiebolaget LM Ericsson (publ)Inventors: Patrick A. Hosein, Mahesh A. Makhijani
-
Publication number: 20100091739Abstract: Methods and systems for improving handover performance by pre-negotiating channel quality information between a serving base station and a target base before a handoff is initiated in a wireless communication system are described. The target base station may send to the serving base station an assigned location within an uplink subframe for a mobile station requesting a handover to the target base station to transmit channel quality information to the target base station. The serving base station may send this information to the mobile station. The mobile station may transmit channel quality information to the target base station at the assigned location.Type: ApplicationFiled: September 16, 2009Publication date: April 15, 2010Applicant: QUALCOMM INCORPORATEDInventors: Pranav Dayal, Mahesh Makhijani
-
Publication number: 20090316815Abstract: Methods and apparatus are provided for increasing throughput in a wireless communication system by reducing the amount of overhead transmitted to certain user terminals. Overhead due to control information may be reduced for these certain user terminals by selecting a low repetition factor. Overhead may be further reduced for these certain user terminals by selecting a modulation/coding scheme with a higher data rate for transmitting the control information. The selection may be based on channel conditions associated with the user terminals, such as signal-to-interference-plus-noise ratios (SINRs).Type: ApplicationFiled: June 17, 2009Publication date: December 24, 2009Applicant: QUALCOMM INCORPORATEDInventors: Serguei A. Glazko, Matthias Brehler, Je Woo Kim, Mahesh Makhijani
-
Publication number: 20090252118Abstract: Techniques for supporting handoff of terminals between a packet-switched network and a circuit-switched network are described. In an aspect, handoff between packet-switched and circuit-switched networks may be facilitated by a designated network entity in the packet-switched network. The designated network entity may interface with both the packet-switched network and the circuit-switched network, perform circuit-switched call origination, and perform handoff procedure. In one design, a first terminal may communicate with the packet-switched network for a packet-switched call with a second terminal. The first terminal may initiate handoff to the circuit-switched network via the designated network entity. The first terminal may perform handoff from the packet-switched network to the circuit-switched network based on an inter-MSC handoff procedure. The first terminal may then communicate with the circuit-switched network for the circuit-switched call with the second terminal after the handoff.Type: ApplicationFiled: April 1, 2009Publication date: October 8, 2009Applicant: QUALCOMM IncorporatedInventors: John Wallace Nasielski, Mahesh A. Makhijani, Arungundram C. Mahendran, Haipeng Jin
-
Publication number: 20090129332Abstract: Providing for an efficient frame structure for wireless communication is disclosed, which may include frequency division multiplexing (FDM) first and second air interfaces in a downlink subframe of a frame, and may also include time division multiplexing (TDM) the first and second air interfaces in an uplink subframe of the frame.Type: ApplicationFiled: October 28, 2008Publication date: May 21, 2009Applicant: QUALCOMM INCORPORATEDInventors: Pranav Dayal, Ayman Fawzy Naguib, Mahesh Makhijani
-
Publication number: 20090075666Abstract: Techniques for sending and receiving signaling messages on shared and dedicated channels in a wireless communication system are described. A user equipment (UE) may receive a first signaling message on a shared channel during a first time period, receive a second signaling message on a dedicated channel during a second time period, and receive a third signaling message on the shared channel during a third time period. The UE may receive traffic data on the shared channel during all time periods. The UE may be handed over from a source Node B to a target Node B during the second time period. The UE may receive the first signaling message via the source Node B, receive the second signaling message via the source and target Node Bs, and receive the third signaling message via the target Node B.Type: ApplicationFiled: September 15, 2008Publication date: March 19, 2009Applicant: QUALCOMM IncorporatedInventors: Mahesh Makhijani, Mehmet Yavuz
-
Publication number: 20080144572Abstract: Carrier frequency selection is enabled in transmitting packet data between a radio transceiver and a mobile station and in a wireless communication network with shared packet data service. The packet data service permits transmission of packet data along a packet data link comprising a plurality of shared packet data channels, such as that provided by the N×EV-DO protocol. Each channel is characterized by a unique carrier frequency. Packet data may be divided and transmitted among the plurality of channels. Packet data may be transmitted along a first of the plurality of channels. Upon receiving a negative acknowledgement indicative of an erroneous packet transmission in accordance with known error control schemes such as HARQ, retransmission data may be transmitted along a second of the plurality of channels. As multiple carrier frequencies are used to transmit packet data and retransmission data, frequency diversity is added to time diversity inherent in retransmission protocols.Type: ApplicationFiled: December 15, 2006Publication date: June 19, 2008Inventor: Mahesh A. Makhijani
-
Patent number: 7346011Abstract: A wireless communication network has radio resources for use to support calls over circuit switched (CS) and packet switched (PS) channels. The network uses such radio resources to first support CS calls and thereafter uses the leftover radio resources to support PS calls. The present invention provides method and apparatus that effectively schedule and support incoming CS calls based on frame-offset selection so as to maximize the leftover radio resources that could be used to support PS calls.Type: GrantFiled: April 28, 2005Date of Patent: March 18, 2008Assignee: Telefonaktiebolaget LM Ericsson (publ)Inventors: Mahesh Makhijani, Srinivasan Balasubramanian, Dang M. Vo