公开(公告)号：GB2603391A

公开(公告)日：2022-08-03

申请号：GB202204966

申请日：2020-09-21

Applicant: IBM

Inventor： PIN-YU CHEN ,  PAYEL DAS ,  RAMAMURTHY KARTHIKEYAN NATESAN ,  PU ZHAO

IPC: G06N3/08 ,  G06F21/00

Abstract: Techniques for sanitization of machine learning (ML) models are provided. A first ML model is received, along with clean training data. A path is trained between the first ML model and a second ML model using the clean training data. A sanitized ML model is generated based on at least one point on the trained path. One or more ML functionalities are then facilitated using the sanitized ML model.

公开(公告)号：GB2579303B

公开(公告)日：2021-02-03

申请号：GB202001676

申请日：2018-07-06

Applicant: IBM

Inventor： PAYEL DAS ,  RYAN HUBBARD

IPC: A61B5/00 ,  A61B5/145 ,  A61B5/24 ,  A61B5/369

Abstract: A method and system of removing artifacts in electroencephalogram (EEG) signals is provided. Digital EEG data, based on a digitizing of analog EEG signals of a patient from a first electrode, is received. Digital sweat sensor data, based on a digitizing of analog sweat sensor signals of the patient that are contemporaneous with the analog EEG signals, is received. A transform is applied to the digital sweat sensor data based on a predetermined sweat stress profile of the patient. The digital EEG data is adaptively adjusted by subtracting the transformed sweat sensor data from the digital EEG data.

公开(公告)号：GB2579303A

公开(公告)日：2020-06-17

申请号：GB202001676

申请日：2018-07-06

Applicant: IBM

Inventor： PAYEL DAS ,  RYAN HUBBARD

IPC: A61B5/00 ,  A61B5/04 ,  A61B5/0476 ,  A61B5/145

Abstract: A method and system of removing artifacts in electroencephalogram (EEG) signals is provided. Digital EEG data, based on a digitizing of analog EEG signals of a patient from a first electrode, is received. Digital sweat sensor data, based on a digitizing of analog sweat sensor signals of the patient that are contemporaneous with the analog EEG signals, is received. A transform is applied to the digital sweat sensor data based on a predetermined sweat stress profile of the patient. The digital EEG data is adaptively adjusted by subtracting the transformed sweat sensor data from the digital EEG data.