Patents by Inventor Pavitra Krishnaswamy
Pavitra Krishnaswamy 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: 20220414464Abstract: There is provided a method of federated machine learning using at least one processor, the method including: transmitting a current global machine learning model to each of a plurality of data sources; receiving a plurality of training updates from the plurality of data sources, respectively, each of the plurality of training updates being generated by the respective data source in response to the global machine learning model received; and updating the current global machine learning model based on the plurality of training updates received and a plurality of data quality parameters associated with the plurality of data sources, respectively, to generate an updated global machine learning model. There is also provided a corresponding server for federated machine learning.Type: ApplicationFiled: December 10, 2019Publication date: December 29, 2022Inventors: Pavitra Krishnaswamy, Lakshminarayanan Anantharaman, Feri Guretno, Mien Ho
-
Patent number: 10716485Abstract: A method for non-invasively resolving electrophysiological activity in sub-cortical structures located deep in the brain by comparing amplitude-insensitive M/EEG field patterns arising from activity in subcortical and cortical sources under physiologically relevant sparse constraints is disclosed. The method includes a sparse inverse solution for M/EEG subcortical source modeling. Specifically, the method employs a subspace-pursuit algorithm rooted in compressive sampling theory, performs a hierarchical search for sparse subcortical and cortical sources underlying the measurement, and estimates millisecond-scale currents in these sources to explain the data. The method can be used to recover thalamic and brainstem contributions to non-invasive M/EEG data, and to enable non-invasive study of fast timescale dynamical and network phenomena involving widespread regions across the human brain.Type: GrantFiled: November 9, 2015Date of Patent: July 21, 2020Assignees: The General Hospital Corporation, Massachusetts Institute of TechnologyInventors: Pavitra Krishnaswamy, Patrick Purdon, Gabriel Obregon-Henao, Matti Hamalainen, Behtash Babadi
-
Patent number: 10646165Abstract: Systems and a method for reducing periodic artifacts in an electrophysiologic signal are provided. The method includes receiving a time-series electrophysiologic signal acquired from a subject and providing a regression model that defines an interference signal caused by periodic artifacts using a harmonic representation. The method also includes applying the regression model using the time-series electrophysiologic signal to define a cost function and performing an iterative optimization process to estimate regression parameters that minimize the cost function. The method further includes determining, using the regression parameters, the interference signal, and generating a corrected time-series electrophysiologic signal by reducing the interference signal.Type: GrantFiled: April 10, 2014Date of Patent: May 12, 2020Assignee: The General Hospital CorporationInventors: Pavitra Krishnaswamy, Patrick L. Purdon, Emery N. Brown
-
Patent number: 10301587Abstract: Disclosed are methods and systems for subnanosecond rise time high voltage (HV) electric pulse delivery to biological loads. The system includes an imaging device and monitoring apparatus used for bio-photonic studies of pulse induced intracellular effects. The system further features a custom fabricated microscope slide having micro-machined electrodes. A printed circuit board to interface the pulse generator to the micro-machined glass slide having the cell solution is disclosed. An low-parasitic electronic setup to interface with avalanche transistor-switched pulse generation system is also disclosed. The pc-board and the slide are configured to match the output impedance of the pulse generator which minimizes reflection back into the pulse generator, and minimizes distortion of the pulse shape and pulse parameters. The pc-board further includes a high bandwidth voltage divider for real-time monitoring of pulses delivered to the cell solutions.Type: GrantFiled: October 31, 2016Date of Patent: May 28, 2019Assignee: UNIVERSITY OF SOUTHERN CALIFORNIAInventors: Pavitra Krishnaswamy, Andras Kuthi
-
Patent number: 9975249Abstract: A neuromuscular model-based controller for a robotic limb having at least one joint includes a neuromuscular model having a muscle model, muscle geometry and reflex feedback loop to determine at least one torque or impedance command to be sent to the robotic limb. One or more parameters that determine relation between feedback data and activation of the muscle model are adjusted consequent to sensory data from at least one of an intrinsic sensor and an extrinsic sensor. A controller in communication with the neuromuscular model is configured to receive the at least one torque or impedance command and controls at least one of position, torque and impedance of the robotic limb joint.Type: GrantFiled: December 22, 2015Date of Patent: May 22, 2018Assignee: Massachusetts Institute of TechnologyInventors: Hugh M. Herr, Ken Endo, Pavitra Krishnaswamy, Jared Markowitz, Michael Frederick Eilenberg, Jing Wang
-
Publication number: 20170332933Abstract: A method for non-invasively resolving electrophysiological activity in sub-cortical structures located deep in the brain by comparing amplitude-insensitive M/EEG field patterns arising from activity in subcortical and cortical sources under physiologically relevant sparse constraints is disclosed. The method includes a sparse inverse solution for M/EEG subcortical source modeling. Specifically, the method employs a subspace-pursuit algorithm rooted in compressive sampling theory, performs a hierarchical search for sparse subcortical and cortical sources underlying the measurement, and estimates millisecond-scale currents in these sources to explain the data. The method can be used to recover thalamic and brainstem contributions to non-invasive M/EEG data, and to enable non-invasive study of fast timescale dynamical and network phenomena involving widespread regions across the human brain.Type: ApplicationFiled: November 9, 2015Publication date: November 23, 2017Inventors: Pavitra Krishnaswamy, Patrick Purdon, Gabriel Obregon-Henao, Matti Hamalainen, Behtash Babadi
-
Publication number: 20170096630Abstract: Disclosed are methods and systems for subnanosecond rise time high voltage (HV) electric pulse delivery to biological loads. The system includes an imaging device and monitoring apparatus used for bio-photonic studies of pulse induced intracellular effects. The system further features a custom fabricated microscope slide having micro-machined electrodes. A printed circuit board to interface the pulse generator to the micro-machined glass slide having the cell solution is disclosed. An low-parasitic electronic setup to interface with avalanche transistor-switched pulse generation system is also disclosed. The pc-board and the slide are configured to match the output impedance of the pulse generator which minimizes reflection back into the pulse generator, and minimizes distortion of the pulse shape and pulse parameters. The pc-board further includes a high bandwidth voltage divider for real-time monitoring of pulses delivered to the cell solutions.Type: ApplicationFiled: October 31, 2016Publication date: April 6, 2017Applicant: University of Southern CaliforniaInventors: Pavitra Krishnaswamy, Andras Kuthi
-
Patent number: 9493765Abstract: Disclosed are methods and systems for subnanosecond rise time high voltage (HV) electric pulse delivery to biological loads. The system includes an imaging device and monitoring apparatus used for bio-photonic studies of pulse induced intracellular effects. The system further features a custom fabricated microscope slide having micro-machined electrodes. A printed circuit board to interface the pulse generator to the micro-machined glass slide having the cell solution is disclosed. An low-parasitic electronic setup to interface with avalanche transistor-switched pulse generation system is also disclosed. The pc-board and the slide are configured to match the output impedance of the pulse generator which minimizes reflection back into the pulse generator, and minimizes distortion of the pulse shape and pulse parameters. The pc-board further includes a high bandwidth voltage divider for real-time monitoring of pulses delivered to the cell solutions.Type: GrantFiled: March 24, 2008Date of Patent: November 15, 2016Assignee: UNIVERSITY OF SOUTHERN CALIFORNIAInventors: Pavitra Krishnaswamy, Andras Kuthi
-
Publication number: 20160207201Abstract: A neuromuscular model-based controller for a robotic limb having at least one joint includes a neuromuscular model having a muscle model, muscle geometry and reflex feedback loop to determine at least one torque or impedance command to be sent to the robotic limb. One or more parameters that determine relation between feedback data and activation of the muscle model are adjusted consequent to sensory data from at least one of an intrinsic sensor and an extrinsic sensor. A controller in communication with the neuromuscular model is configured to receive the at least one torque or impedance command and controls at least one of position, torque and impedance of the robotic limb joint.Type: ApplicationFiled: December 22, 2015Publication date: July 21, 2016Inventors: Hugh M. Herr, Ken Endo, Pavitra Krishnaswamy, Jared Markowitz, Michael Frederick Eilenberg, Jing Wang
-
Publication number: 20160038091Abstract: Systems and a method for reducing periodic artifacts in an electrophysiologic signal are provided. The method includes receiving a time-series electrophysiologic signal acquired from a subject and providing a regression model that defines an interference signal caused by periodic artifacts using a harmonic representation. The method also includes applying the regression model using the time-series electrophysiologic signal to define a cost function and performing an iterative optimization process to estimate regression parameters that minimize the cost function. The method further includes determining, using the regression parameters, the interference signal, and generating a corrected time-series electrophysiologic signal by reducing the interference signal.Type: ApplicationFiled: April 10, 2014Publication date: February 11, 2016Applicant: THE GENERAL HOSPITAL CORPORATIONInventors: Pavitra Krishnaswamy, Patrick L. Purdon
-
Patent number: 9221177Abstract: A neuromuscular model-based controller for a robotic limb having at least one joint includes a neuromuscular model having a muscle model, muscle geometry and reflex feedback loop to determine at least one torque or impedance command to be sent to the robotic limb. One or more parameters that determine relation between feedback data and activation of the muscle model are adjusted consequent to sensory data from at least one of an intrinsic sensor and an extrinsic sensor. A controller in communication with the neuromuscular model is configured to receive the at least one torque or impedance command and controls at least one of position, torque and impedance of the robotic limb joint.Type: GrantFiled: April 18, 2013Date of Patent: December 29, 2015Assignee: Massachusetts Institute of TechnologyInventors: Hugh M. Herr, Ken Endo, Pavitra Krishnaswamy, Jared Markowitz, Michael Frederick Eilenberg, Jing Wang
-
Publication number: 20130310979Abstract: A neuromuscular model-based controller for a robotic limb having at least one joint includes a neuromuscular model having a muscle model, muscle geometry and reflex feedback loop to determine at least one torque or impedance command to be sent to the robotic limb. One or more parameters that determine relation between feedback data and activation of the muscle model are adjusted consequent to sensory data from at least one of an intrinsic sensor and an extrinsic sensor. A controller in communication with the neuromuscular model is configured to receive the at least one torque or impedance command and controls at least one of position, torque and impedance of the robotic limb joint.Type: ApplicationFiled: April 18, 2013Publication date: November 21, 2013Applicant: Massachusetts Institute of TechnologyInventors: Hugh M. Herr, Ken Endo, Pavitra Krishnaswamy, Jared Markowitz, Michael Frederick Eilenberg, Jing Wang
-
Publication number: 20080231337Abstract: Disclosed are methods and systems for subnanosecond rise time high voltage (HV) electric pulse delivery to biological loads. The system includes an imaging device and monitoring apparatus used for bio-photonic studies of pulse induced intracellular effects. The system further features custom fabricated microscope slide having micro-machined electrodes. A printed circuit board to interface the pulse generator to the micro-machined glass slide having the cell solution. An low-parasitic electronic setup to interface with avalanche transistor-switched pulse generation system. The pc-board and the slide are configured to match the output impedance of the pulse generator which minimizes reflection back into the pulse generator, and minimizes distortion of the pulse shape and pulse parameters. The pc-board further includes a high bandwidth voltage divider for real-time monitoring of pulses delivered to the cell solutions.Type: ApplicationFiled: March 24, 2008Publication date: September 25, 2008Inventors: Pavitra Krishnaswamy, Andras Kuthi
-
Publication number: 20050050211Abstract: A method and apparatus to manage network addresses are described.Type: ApplicationFiled: August 29, 2003Publication date: March 3, 2005Inventors: Bharat Kaul, Narendra Tulpule, Mingcheng Zhu, Pavitra Krishnaswamy